New Developments of IT, IoT and ICT Applied to Agriculture: Proceedings of ICAIT 2019 [1st ed.] 9789811550720, 9789811550737

Table of contents :
Front Matter ....Pages i-xvi
Front Matter ....Pages 1-1
Vegetable Distribution System Using Machine Learning Approach (Ari Aharari, Chunsheng Yang, Jair Minoro Abe)....Pages 3-9
A Purpose of a Smart Vegetable Garden Model Based on Paraconsistent Annotated Evidential Logic Eτ (Jonatas Santos de Souza, Jair Minoro Abe, Luiz Antônio de Lima, Kazumi Nakamatsu)....Pages 11-18
Paraconsistent Artificial Neural Network Applied to Agribusiness (Taciana Tamyris Alves de Souza, Cristina Corrêa de Oliveira, Jair Minoro Abe, Ari Aharari, Kazumi Nakamatsu)....Pages 19-28
Front Matter ....Pages 29-29
Research and Design on an Automatic Control System Using in Antarctic Greenhouse (Kaiyan Lin, Chang Liu, Jie Chen, Junhui Wu, Huiping Si)....Pages 31-45
Design of Microwave Triple-Frequency Multimeter Based on Multilayer MoS2 (Zhilong Zhao, Xiaoling Zhong, Ting Ting Guo, Bing Wu, Li Wen, Liangyi Deng et al.)....Pages 47-54
Correlation Research on the Structure of the Apple Tree Vigor and Its Fruit Quality (Zhijun Wang, Peng Lan, Fenggang Sun)....Pages 55-63
Real-Time Data Transmission Design of Unmanned Aerial Vehicle Gamma Spectrometer (Xiaoling Zhong, ZhiLong Zhao, Yuting Jiang, Bing Wu, MingHao Guo, Yong Fang)....Pages 65-75
0.8 GHz Low-Noise Amplifier Design (Xueshi Hou, Xiaoling Zhong, Zhilong Zhao, Liangyi Deng, Han Mei, Xue Wei et al.)....Pages 77-87
Optimization Analysis of Laying Length of Thin-Wall Micro-Spray Belt Under Different Slope Conditions (Jin Yi Wang, Lu Hua Yang, Wan Li Gou, Yan Hui Dong)....Pages 89-97
L-Band Ultra-Wideband Low-Noise Amplifier Design (Xiaoling Zhong, Haoxuan Sheng, Yong Fang, Yong Guo, Baiqiu Liu, Zhilong Zhao et al.)....Pages 99-108
Research on Pipe Surface Defect Recognition Based on Convolutional Neural Network (Zehui Yuan, Hui Guo, Shaoping Zhou)....Pages 109-118
Research on the Method of Tea Disease Recognition Based on Deep Learning (Tengyue Mao, Feng Liu, Bo Huang, Liuqiang Wang)....Pages 119-128
A Research on Tea Traceability Consensus Mechanism Based on Blockchain Technology (Tengyue Mao, Ying Fan, Juan Yang, Hengbin Wei)....Pages 129-137
The Research on the Judgment Method for Porcine Abnormal Diet Based on Improved PSO-SVDD (Sunan Zhang, Jianyan Tian, Jiangli Li)....Pages 139-146
The Design of Two-Channel Outputs Switching Mode Power Supply Based on TOP100Y (Cao Yong, Zheng Yifan, Wang Xiao, Zhou Heng, Liu Yi)....Pages 147-155
Design of Data Acquisition System for ASA Test (Cao Yong, Zheng Yifan, Wang Xiao, Zhou Heng, Liu Yi)....Pages 157-166
Raman Perception Monitoring Application of IoT Technology in Agricultural Planting (Wu Yin, Chenying He)....Pages 167-176
Visible and Infrared Image Fusion Based on Masked Online Convolutional Dictionary Learning with Frequency Domain Computation (Chengfang Zhang, Xingchun Yang)....Pages 177-182
Review and Optimization of Agricultural Support Protection Subsidy Policy (Yingjun Xie, Gaoxiu Liu)....Pages 183-190
Intelligent Greenhouse Information Collection and Control System Based on Internet of Things (Wei Han, Pingzeng Liu, Jianyong Zhang, Jianhang Fu, Yuting Yu, Xianglong Wang et al.)....Pages 191-199
Design and Key Technology Research of Civil-Military Integration Regulations and Standards Management System (Tianming Huang)....Pages 201-207
A Reliable Wireless Monitor and Control System with Low Power for Greenhouse Microclimate (Zhenfeng Xu, Jingjing Yin, Xiujuan Li)....Pages 209-216
Prediction of Excessive Cadmium in Rice Based on Weighted Bayesian Fusion Model (Baohua Zhang, Wei Wang, Yi An, Yuan Jiao, Yue Li)....Pages 217-225
Segmentation of Oilseed Rape Flowers Based on HSI Color Space and Local Region Clustering (Jiahua Zeng, Xuan Wang, Kaiqiong Sun)....Pages 227-232
Agricultural Remote Monitoring and Alarm System Based on LoRa and Internet of Things Cloud Platform (TingTing Chen, XinChen Zhang, BoWen Zhao)....Pages 233-239
Aerial Triangulation Study of Unmanned Aerial Vehicles in Forest Area Based on Pix4D (Jing Zhao, Jian Guan, Yan-jie Li, Hong-wei Du, Zhi-min Zhu, Yue Zhao et al.)....Pages 241-248
Back Matter ....Pages 249-250

Citation preview

Smart Innovation, Systems and Technologies 183

Kazumi Nakamatsu · Roumen Kountchev · Ari Aharari · Nashwa El-Bendary · Bin Hu   Editors

New Developments of IT, IoT and ICT Applied to Agriculture Proceedings of ICAIT 2019

123

Smart Innovation, Systems and Technologies Volume 183

Series Editors Robert J. Howlett, Bournemouth University and KES International, Shoreham-by-sea, UK Lakhmi C. Jain, Faculty of Engineering and Information Technology, Centre for Artificial Intelligence, University of Technology Sydney, Sydney, NSW, Australia

The Smart Innovation, Systems and Technologies book series encompasses the topics of knowledge, intelligence, innovation and sustainability. The aim of the series is to make available a platform for the publication of books on all aspects of single and multi-disciplinary research on these themes in order to make the latest results available in a readily-accessible form. Volumes on interdisciplinary research combining two or more of these areas is particularly sought. The series covers systems and paradigms that employ knowledge and intelligence in a broad sense. Its scope is systems having embedded knowledge and intelligence, which may be applied to the solution of world problems in industry, the environment and the community. It also focusses on the knowledge-transfer methodologies and innovation strategies employed to make this happen effectively. The combination of intelligent systems tools and a broad range of applications introduces a need for a synergy of disciplines from science, technology, business and the humanities. The series will include conference proceedings, edited collections, monographs, handbooks, reference books, and other relevant types of book in areas of science and technology where smart systems and technologies can offer innovative solutions. High quality content is an essential feature for all book proposals accepted for the series. It is expected that editors of all accepted volumes will ensure that contributions are subjected to an appropriate level of reviewing process and adhere to KES quality principles. ** Indexing: The books of this series are submitted to ISI Proceedings, EI-Compendex, SCOPUS, Google Scholar and Springerlink **

More information about this series at http://www.springer.com/series/8767

Kazumi Nakamatsu Roumen Kountchev Ari Aharari Nashwa El-Bendary Bin Hu •

•

•

•

Editors

New Developments of IT, IoT and ICT Applied to Agriculture Proceedings of ICAIT 2019

123

Editors Kazumi Nakamatsu School of HSE University of Hyogo Himeji, Hyogo, Japan Ari Aharari Faculty of Computer and Information Sciences Sojo University Nishi-ku, Kumamoto, Japan Bin Hu China Mobile Communications Corporation Xiangtan, Hunan, China

Roumen Kountchev Department of Radiocommunications and Video Technologies Technical University of Sofia Sofia, Bulgaria Nashwa El-Bendary China-Arab States Technology Transfer Center Arab Academy for Science, Technology, and Maritime Transport (AASTMT) Gîza, Egypt

ISSN 2190-3018 ISSN 2190-3026 (electronic) Smart Innovation, Systems and Technologies ISBN 978-981-15-5072-0 ISBN 978-981-15-5073-7 (eBook) https://doi.org/10.1007/978-981-15-5073-7 © Springer Nature Singapore Pte Ltd. 2021 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

ICAIT-2019 Organization

Honorary Chair: Prof. Lakhmi Jain, Australia General Chair: Prof. Kazumi Nakamatsu, Japan Conference Co-chairs: Assoc. Prof. Ari Aharari, Japan Assoc. Prof. Nashwa El-Bendary, Egypt International Advisory Board Chair: Prof. Srikanta Patnaik, India Program Chair: Prof. Roumen Kountchev, Romania Organizing Chair: Mr. Silai Zhou, China

ICAIT2019 International Advisory Board and Program Committee Roumen Kountchev, Bulgaria Jair M. Abe, Brazil Fabio Romeu de Carvalho, Brazil Ari Aharari, Japan Nashwa El-Bendary, Egypt Aboul Ela Hassanian, Egypt Tulay Yildirim, Turkey Altas Ismail, Turkey Valentina E. Balas, Romania

v

vi

Margarita Favorskaya, Russia Andrei Tyugashev, Russia Ajith Abraham, USA Vincenzo Piuri, Italy Morgado Dias, Portugal Nicolae Paraschiv, Romania Sachio Hirokawa, Japan Siddhartha Bhattacharyya, India Ladjel Bellatreche, France Juan D. Velasquez Silva, Chile Hossam A. Gaber, Canada Ngoc Thanh Nguyen, Poland Mario Divan, Argentina Petia Koprinkova, Bulgaria Jawad K. Ali, Iraq Sufyan T. Faraj, Iraq Hercules A. Prado, Brazil Kiyota Hashimoto, Thailand Florin P. Vladicescu, Romania Lorna Uden, UK Anne Hakansson, Norway Sunil Kumar Khatri, India Mabrouk Omrani, USA Rumen Arnoudov, Bulgaria Ding Li Ya, Japan Tru Cao, Vietnam Triet Tran, Vietnam Tho Quan, Vietnam Van Du Nguyen, Japan Minh Le Nguyen, Japan Bogdan Trwawinski, Poland Pawel Sitek, Poland Rumen Mironov, Bulgaria Veska Georgieva, Bulgaria Ruminia Kounchev, Bulgaria Toshifumi Kimura, Japan Debanjan Konar, India Sandip Dey, India Sourav De, India Recep Cakmak, Turkey Marius Olteanu, Romania Aslina Baharum, Malaysia Tung Nguyen, Vietnam Horacio Leone, Argentina Ricardo Medel, Argentina

ICAIT-2019 Organization

ICAIT-2019 Organization

Marcelo Marciszack, Argentina Alena Favorskaya, Russia Kleber X. Sampaio de Souza, Brazil Aluízio Haendchen Filho, Brazil Edilson Ferneda, Brazil Recep Cakmak, Turkey Christoph Ruland, Germany Qusay Mahmoud, Canada Marius Balas, Romania Simona Dzitac, Romania Emil Pricop, Romania Otilia Cangea, Romania Mihaela Luca, Romania Dmitry Zaitsev, Ukraine Sergei Prokhorov, Russia Dmitry Ivanov, Russia Sergei Orlov, Russia Boris Sokolov, Russia Mikhail Sergeev, Russia Xilong Qu, China Hao Wang, China Liang Zong, China Junjie Lv, China Xilang Tang, China Organizing Institution IRNet International Academy Communication Center, Wuhan, China

vii

Preface

The 1st International Conference on Agriculture and IT/IoT/ICT (ICAIT2019) has been held in Wuhan, Hubei, China during October 18–20, 2019. If we consider the food problem in an aging society, especially aging issue of farmers, one of the most important subjects is increasing the productivity of agricultural produce by technical innovation in agriculture. Actually, collaboration of information technologies and various parts of agriculture has become very important such as sensor networks in green houses recently. The following three domains are nominated as such technical innovation can be expected to be developed and effective: domains subject to (1) agriculture product itself such as improvement of species, (2) agricultural produce environment such as cultivation techniques and infrastructure of farms, and (3) distribution and selling of agricultural products such as transportation and preservation. Furthermore, a lot of information technologies have already been developed theoretically and practically, and utilized in various farms recently. Many farmers want to introduce IT to their farms in the future though most of the utilization of IT at farms are information collection by Internet or saving accounting information of farm management, and the utilization of IT directly connecting with increasing agricultural productivity is quite few. This unbalance is due to lack of knowledge of IT. The First International Conference on Agriculture and Information Technologies (IT)/Internet Communication Technologies (ICT)/Internet of Things (IoT) ICAIT2019 provided a common platform for exchanging ideas of all kinds of information technologies such as Information and Communication Technologies (ICT) and Internet of Things (IoT) applicable to various parts of agriculture and its infrastructure by academic researchers, students who are studying foundation or application of IT and agriculture, people concerned with IT-based agricultural business, farmers considering to introduce IT to their farms, etc. Therefore, the

ix

x

Preface

scope of the conference covers very wide topics in agriculture and information technologies as follows: Recent Advances in Intelligent Paradigms for Information Technologies, Information Technologies Applicable to Smart Agriculture, Intelligent Information Systems for Smart Farm Systems, Web based Intelligent Systems on Agriculture, ICT based Marketing of Agricultural Products, IoT for Agricultural Produce and Products, Soft Computing Theories Applied to Agricultural Technology, Intelligent Systems for Production Engineering applied to Agriculture, Hybrid Intelligent Systems for Bioinformatics, Agricultural Product Consumption Network Systems, Network Systems for Agricultural Produce, Network Systems for Agricultural Food Processing, Soft Computing Techniques for Smart Agriculture, Data Science for Agriculture, Intelligent Paradigm applicable to Agricultural Food Science, IoT based Robotics and Automation for Agricultural Produce, Intelligent Logistics for Agriculture, Intelligent Control for Infrastructure of Agriculture, Intelligent Control for Agricultural Machines, Machine Learning for Agriculture, and Intelligent Agricultural Management. We accepted 3 invited and 23 regular papers among submitted 43 papers from China, India, Japan, Brazil, etc. at ICAIT2019. This volume is devoted to presenting all those accepted papers of ICAIT2019. Lastly, we wish to express our sincere appreciation to all individuals and organizations who have contributed to ICAIT2019, especially to our colleagues in the program committee for their thorough review of all the submissions, which is vital to the success of ICAIT2019, and also to the members in the organizing committee who had dedicated their time and efforts in planning, promoting, organizing, and helping the conference. Special appreciation is extended to our plenary speakers: Prof. Jair M. Abe, from Paulista University, Sao Paulo, Brazil and Prof. Andrey A. Tyugashev, from Samara State University of Transport, Samara, Russia, who made speeches titled Contemporary Non-classical Approaches to Artificial General Intelligence and Consistent Intelligent Real-Time Control of Complex Systems, respectively. Xi’an, Shanxi, China November 2019

Kazumi Nakamatsu Roumen Kountchev Ari Aharari Nashwa El-Bendary Bin Hu

Contents

Part I 1

2

3

Vegetable Distribution System Using Machine Learning Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ari Aharari, Chunsheng Yang, and Jair Minoro Abe A Purpose of a Smart Vegetable Garden Model Based on Paraconsistent Annotated Evidential Logic Es . . . . . . . . . . . . . . Jonatas Santos de Souza, Jair Minoro Abe, Luiz Antônio de Lima, and Kazumi Nakamatsu Paraconsistent Artificial Neural Network Applied to Agribusiness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Taciana Tamyris Alves de Souza, Cristina Corrêa de Oliveira, Jair Minoro Abe, Ari Aharari, and Kazumi Nakamatsu

Part II 4

5

6

Invited Papers 3

11

19

Regular Papers

Research and Design on an Automatic Control System Using in Antarctic Greenhouse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kaiyan Lin, Chang Liu, Jie Chen, Junhui Wu, and Huiping Si Design of Microwave Triple-Frequency Multimeter Based on Multilayer MoS2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zhilong Zhao, Xiaoling Zhong, Ting Ting Guo, Bing Wu, Li Wen, Liangyi Deng, and Yuting Jiang Correlation Research on the Structure of the Apple Tree Vigor and Its Fruit Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zhijun Wang, Peng Lan, and Fenggang Sun

31

47

55

xi

xii

7

Contents

Real-Time Data Transmission Design of Unmanned Aerial Vehicle Gamma Spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiaoling Zhong, ZhiLong Zhao, Yuting Jiang, Bing Wu, MingHao Guo, and Yong Fang

8

0.8 GHz Low-Noise Amplifier Design . . . . . . . . . . . . . . . . . . . . . . . Xueshi Hou, Xiaoling Zhong, Zhilong Zhao, Liangyi Deng, Han Mei, Xue Wei, Yuting Jiang, Baiqiu Liu, and Yong Fang

9

Optimization Analysis of Laying Length of Thin-Wall Micro-Spray Belt Under Different Slope Conditions . . . . . . . . . . . . Jin Yi Wang, Lu Hua Yang, Wan Li Gou, and Yan Hui Dong

10 L-Band Ultra-Wideband Low-Noise Amplifier Design . . . . . . . . . . Xiaoling Zhong, Haoxuan Sheng, Yong Fang, Yong Guo, Baiqiu Liu, Zhilong Zhao, and Yangyang Wang

65

77

89 99

11 Research on Pipe Surface Defect Recognition Based on Convolutional Neural Network . . . . . . . . . . . . . . . . . . . . 109 Zehui Yuan, Hui Guo, and Shaoping Zhou 12 Research on the Method of Tea Disease Recognition Based on Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 Tengyue Mao, Feng Liu, Bo Huang, and Liuqiang Wang 13 A Research on Tea Traceability Consensus Mechanism Based on Blockchain Technology . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Tengyue Mao, Ying Fan, Juan Yang, and Hengbin Wei 14 The Research on the Judgment Method for Porcine Abnormal Diet Based on Improved PSO-SVDD . . . . . . . . . . . . . . . . . . . . . . . 139 Sunan Zhang, Jianyan Tian, and Jiangli Li 15 The Design of Two-Channel Outputs Switching Mode Power Supply Based on TOP100Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 Cao Yong, Zheng Yifan, Wang Xiao, Zhou Heng, and Liu Yi 16 Design of Data Acquisition System for ASA Test . . . . . . . . . . . . . . 157 Cao Yong, Zheng Yifan, Wang Xiao, Zhou Heng, and Liu Yi 17 Raman Perception Monitoring Application of IoT Technology in Agricultural Planting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Wu Yin and Chenying He 18 Visible and Infrared Image Fusion Based on Masked Online Convolutional Dictionary Learning with Frequency Domain Computation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 Chengfang Zhang and Xingchun Yang

Contents

xiii

19 Review and Optimization of Agricultural Support Protection Subsidy Policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 Yingjun Xie and Gaoxiu Liu 20 Intelligent Greenhouse Information Collection and Control System Based on Internet of Things . . . . . . . . . . . . . . . . . . . . . . . . 191 Wei Han, Pingzeng Liu, Jianyong Zhang, Jianhang Fu, Yuting Yu, Xianglong Wang, Lin Xu, and Ningning Cui 21 Design and Key Technology Research of Civil-Military Integration Regulations and Standards Management System . . . . . 201 Tianming Huang 22 A Reliable Wireless Monitor and Control System with Low Power for Greenhouse Microclimate . . . . . . . . . . . . . . . . . . . . . . . . 209 Zhenfeng Xu, Jingjing Yin, and Xiujuan Li 23 Prediction of Excessive Cadmium in Rice Based on Weighted Bayesian Fusion Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 Baohua Zhang, Wei Wang, Yi An, Yuan Jiao, and Yue Li 24 Segmentation of Oilseed Rape Flowers Based on HSI Color Space and Local Region Clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Jiahua Zeng, Xuan Wang, and Kaiqiong Sun 25 Agricultural Remote Monitoring and Alarm System Based on LoRa and Internet of Things Cloud Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 TingTing Chen, XinChen Zhang, and BoWen Zhao 26 Aerial Triangulation Study of Unmanned Aerial Vehicles in Forest Area Based on Pix4D . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 Jing Zhao, Jian Guan, Yan-jie Li, Hong-wei Du, Zhi-min Zhu, Yue Zhao, and Teng-fei Ma Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

About the Editors

Kazumi Nakamatsu received the Ms. Eng. and Dr. Sci. from Shizuoka University, 1976 and Kyushu University, Japan, 1999, respectively. He is a full Professor at School of Human Science and Environment, University of Hyogo, Japan. His research interests encompass various kinds of logic and their applications to Computer Science, especially paraconsistent annotated logic programs and their applications. He is an author of over 150 journal papers, book chapters and conference papers, and edited 14 books published by prominent publishers such as Springer-Verlag. He has chaired various international conferences, workshops and invited sessions, and he has been a member of numerous international program committees of workshops and conferences in the area of Artificial Intelligence and Computer Science. He serves as the Founding-Editor and Editor-in-Chief of the International Journal of Reasoning-based Intelligent Systems by Inderscience Publishers (UK), and an editorial board member of many international journals. He has contributed numerous invited lectures at international workshops, conferences, and academic organizations. He also is a recipient of some conference and paper awards. He is a member of Japan AI Society, etc. Prof. Dr. Roumen Kountchev, D.Sc. is with the Faculty of Telecommunications, Dept. of Radio Communications and Video Technologies - Technical University of Sofia, Bulgaria. He has 341 papers published in magazines and conference proceedings, 15 books, 46 book chapters; 20 patents. A member of Euro Mediterranean Academy of Arts and Sciences; President of Bulgarian Association for Pattern Recognition (member of IAPR); Editorial board member of IJBST Journal Group; Editorial board member of: Intern. J. of Reasoning-based Intelligent Systems; Intern. J. Broad Research in Artificial Intelligence and Neuroscience. Assoc. Prof. Ari Aharari He received M.E. and Ph.D. in Industrial Science and Technology Engineering and Robotics from Niigata University and Kyushu Institute of Technology, Japan, in 2004 and 2007, respectively. In 2004, he joined GMD-JAPAN as a Research Assistant. He was Research Scientist and Coordinator at FAIS- Robotics Development Support Office from 2004 to 2007. He was a xv

xvi

About the Editors

Postdoctoral Research Fellow of the Japan Society for the Promotion of Science (JSPS) at Waseda University, Japan, from 2007 to 2008. He served as a Senior Researcher of Fukuoka IST involved in the Japan Cluster Project from 2008 to 2010. In 2010, he became an Assistant Professor at the Faculty of Informatics of Nagasaki Institute of Applied Science. Since 2012, he has been an Associate Professor at the Department of Computer and Information Science, SOJO University, Japan. His research interests are IoT, Robotics, IT Agriculture, Image Processing, and Data Analysis (Big Data) and their applications. He is a member of the IEEE (Robotics and Automation Society), RSJ (Robotics Society of Japan), IEICE (Institute of Electronics, Information and Communication Engineers), IIEEJ (Institute of Image Electronics Engineers of Japan). Dr. Nashwa El-Bendary received her Ph.D. in information technology from the Faculty of Computers and Information, Cairo University, Egypt in 2008. From 2008 to 2015, she was an Assistant Professor with the College of Management and Technology and since 2016, she has been an Associate Professor with the College of Computing and Information Technology at the Arab Academy for Science, Technology, and Maritime Transport (AASTMT). Dr. El-Bendary is the author of many articles published in indexed international journals and conferences. She also participated in several international research projects. Her research interests include image and signal processing, machine learning, pattern recognition, and internet of things. Dr. El-Bendary is an IEEE Senior member and she was a recipient of the UNESCO-ALECSO Award for creativity and technical innovation for young researchers in 2014, and The L'Oréal-UNESCO For Women in Science Fellowship in 2015. She co-organized and co-chaired various special sessions and has been invited as a speaker under the framework of several distinguished international conferences. She is also an editorial board member of several international indexed journals. Bin Hu engineer, a manager of CMCC, Hunan company, he is the co-founder of IRNet International Academic Communication Center (IRNet China), a special delegate of Interscience Research Network (IRNet), he is the guest editor of EI-indexed journals: International Journal of Reasoning-based Intelligent Systems (IJRIS) and International Journal of Information Systems and Supply Chain Management (IJISSCM), the editor board member of International Journal of Enterprise Information Systems (IJEIS) and International Journal of Strategic Decision Sciences (IJSDS) and the technical editor of International Journal of Computer & Communication Technology (IJCCT). Recently, he edited 2 special issues in IJISSCM and 2 special issues in IJRIS, he published many articles in high standard journals like Knowledge-based System (KS), Journal of Intelligent & Fuzzy Control (JIFS), Enterprise Information System (EIS) and etc.

Part I

Invited Papers

Chapter 1

Vegetable Distribution System Using Machine Learning Approach Ari Aharari, Chunsheng Yang, and Jair Minoro Abe

Abstract The aging of the farmer has progressed in Japan rapidly, and successor problem is becoming more serious. Most of the researches have been focused on high quality, mass production, and management of field server. However, the establishment of new product distribution and management technology is a major challenge, especially for new farmers. In this paper, we focus on the vegetable distribution system to solve the management issues of the farming acquisition of technology. We also propose newly designed machine learning based vegetable distribution system which is controlled by market factors, field environmental information, and weather information.

1.1 Introduction In recent years, there is a rapid reduction and aging of the Japanese agriculture workers, and there are no clear solutions for their replacement in the short term. According to the Ministry of Agriculture, Forestry, and Fisheries in 1960, the total agriculture worker’s population was approximately 14.54 million people. Meanwhile, in the 2013 statistics, the total population of agriculture workers is 2.39 million people where 1.478 million people are 65 years old or older, which represents 60%. This leads to the conclusion that the situation has been getting worse in a fast manner [1–3]. Under such circumstances, we observed that around 50,000 people enter as new agriculture workers during the year 2013.

A. Aharari (B) Sojo University, Kumamoto, Japan e-mail: [email protected] C. Yang National Research Council Canada, Ottawa, Canada J. M. Abe Paulista University, São Paulo, Brazil © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_1

3

4

A. Aharari et al.

However, according to the survey results related to the farming situation of new farmers done by the National Agricultural Chamber of the National new farming consultation center, securing farmland for new farmers, securing funds, and the acquisition of farming technology is a major management and distribution issue. In particular, for the acquisition of farming technology, it is pointed out that 55.5% of people who answered the questionnaire experienced a hard time doing this process. For farming technology, important factors related to the healthy growth of the crop are temperature management and water management. Documents related to irrigation information or reference materials published by the Ministry of Agriculture, Forestry, and Fisheries were not updated regularly and were not related to the different soil conditions presented in lands around the different Japanese prefectures or their related water characteristics. Having information related to the soil moisture is relevant for the farmer since it is vital for healthy crop growth. Therefore, it is important that the development of such a system and the related apparatus grasp the different moisture levels in the soil [4, 5]. In other hands, after producing the vegetable, the distribution system which is connected directly to market needs and feedback the important needs and market prices to the farmer is an important research domain. In this study, we focus our efforts to solve the management and distribution issues of the farming acquisition of technology and to develop a vegetable distribution system by applying a machine learning approach.

1.2 Current Technologies Several projects have been done in this research domain to support, especially, new farmers. But most of the systems proposed are initially expensive which make difficulty for farmers of middle or small fields. Hence, almost all of them are focusing to support the farmer till harvesting and they did not approach to give feedback of the market, which may affect directly the farmer’s income.

1.2.1 Fujitsu Akisai Fujitsu’s Akisai system is a SaaS (Software as a Service) product. It is not exclusive for agriculture purposes but cattle as well. Akisai is a management system that integrates data generated in the field through sensor network and mobile devices into a much enterprise-like managerial structure. This structure consolidates functions like business analysis, business administration, production management (agricultural production management, beef production management, cattle husbandry, soil analysis—fertilizer planning and greenhouse environmental control), and Sales (sales management and sales delivery).

1 Vegetable Distribution System Using Machine Learning Approach

5

This system was launched in 2012, and it is designed to integrate a big number of farmers and its main objective is to locate their products in a fair business environment. The system has two options depending on the management level farmers want to allow the system to do. The biggest disadvantage from the point of view of the farmer is the cost. The farmers are financially struggling because they are not able to commercialize the 100% of their produce at a single price since paying a substantial amount of money for this service could represent more a problem than a solution if the system cannot ensure to commercialize the 100% of their produce. Another disadvantage is that farmers have to improve their knowledge regarding some IT tools the system use. In that case, farmers can find troublesome with the system operation.

1.2.2 Toshiba FoodCaster Ex FoodCaster is a sales management service. As we mentioned before, a key factor for the farmer is to successfully commercialize if not possible close to 100% of their produce. In this regard, FoodCaster is based in four main cycle areas: Plan, Do, Check, and Action. The general idea of the system is to seamlessly integrate farmer’s produce and potential markets. The system can create easy to understand charts based on the sales and different sales management-related parameters. The end user needs to input the required information for the charts to be made together with some automated information the system will retrieve from different authorized third party sources. As Fujitsu’s Akisai, the main disadvantage is the price farmers have to pay to take advantage of the different system capabilities.

1.2.3 NEC “IT-Aided Mikan Production” “IT-aided Mikan Production” is a project accepted to promote practical research by the Minister of Agriculture, Forestry, and Fisheries (MAFF) of Japan. The project is being done by different organizations among them are Graduate School of Mie University, Mie Prefecture Agricultural Research Institute, Kumano Agriculture, Forestry, Commerce, Industry and Environment Office, etc. The system’s target is to create a platform where experienced farmers can share their expertise and knowledge with novice or young farmers. The reason this platform was created is to solve the farmer’s aging problem in Japan, where aging farmers cannot find young farmers to share/pass their knowledge to ensure business continuity. This system is still in the research phase, but it mentioned their intentions to launch as a commercial application and increase the number of crops the system will work on. Figure 1.1 shows the system outline.

6

Fig. 1.1 NEC system outline

A. Aharari et al.

1 Vegetable Distribution System Using Machine Learning Approach

7

1.3 Proposed Approach The stable food supply is an important theme that is indispensable for maintaining and promoting health. Therefore, the research and development of a highly profitable farming support system, a next-generation facility horticulture model, to strengthen the food production base are being promoted in Japan. On the other hand, it is difficult to predict and control vegetable harvest time and quantity for farmers who carry out production activities in the natural environment. Therefore, the farm’s business management is stressed by various risks such as disposal and compensation costs. In particular, vegetables are easily lost during mis-optimization, and management of growth prospects and shipping dates is one of the major issues in achieving management stability for farmers. Traditionally, farmers made predictions based on intuition and their experience and made a planned adjustment one month ago with an error of roughly one week [8]. In this research, we propose an approach (Fig. 1.2) that is capable to predict the harvest and shipment time of products using market orientation such as consumers, distribution, and restaurant needs. Multimodal Deep Learning (MDL) is used as a method to predict the harvest, shipment time from environmental data, consumerism,

Fig. 1.2 Proposed approach outline

8

A. Aharari et al.

distribution information, and restaurants with different attributes. The goal is to stabilize farmers’ sales by raising prices and expanding consumption. • Environmental Data: One main factor related to the healthy crop growth is the presence/absence of moisture, amount of water present in the soil, temperature, illuminance, etc. Here, we have developed an affordable system that could perform environmental data in real-time. Besides, the measured data will be stored in a cloud service for future or further [7]. The system needs to be calibrated to provide accurate measures to the client; therefore, a sample data is taken in a controlled environment, and the system is calibrated accordingly. • Distribution Information: In the proposed approach, we applied IoT devices to trace the vegetable distribution from the farm to the restaurant [6]. All daily orders, market price, and stock information are collected to provide the market needs. • Consumer Intention: Consumer purchase history, attribute, and the price in the shop are also other important information that can help us to analyze the market needs. All these information can be collected by Point of Sale System (POS) from local shops. • Restaurant Information: Daily vegetable order, menu, and the POS information from the restaurant also provide valuable information about market needs. • Farmers’ sales stabilization: All the above information are valuable data which are used for analysis in the Multimodal Deep Learning (MDL) in the proposed approach. The feedback from MDL can help the farmers to estimate the best period for a planned adjustment. Because the output from the MDL is looked up from the market needs, it could be stabilizing farmers’ sales by raising prices and expanding consumption.

1.4 Conclusion In this paper, we proposed an approach that is capable to predict the harvest and shipment time of products using market orientation such as consumers, distribution, and restaurant needs. Multimodal Deep Learning (MDL) is applied as a method to predict the harvest, shipment time from environmental data, consumerism, distribution information, and restaurants with different attributes. Future works are to collect the information from the real environment and evaluate the efficiency of proposed approach.

References 1. World Bank: Food Price Watch, February 2011 2. Monthly Statistics of Agriculture: Forestry and Fisheries, Japanese Ministry of Agriculture Forestry and Fisheries, November 2014

1 Vegetable Distribution System Using Machine Learning Approach

9

3. Dennis, A.L.R., Ahrary A.: Big data approach in an ICT agriculture project. In: 5th IEEE International Conference on Awareness Science and Technology (iCAST 2013), Aizu-Wakamatsu, Japan, pp. 261–265 (2013) 4. Dennis, A.L.R., Ahrary, A.: A big data approach for a new ICT agriculture application development. In: 2013 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery (CyberC 2013), Beijing, China, pp. 140–143 (2013) 5. Meeder, B., Tam, J., Gage Kelley, P., Faith Cranor, L.: RT @IWantPrivacy: wide-spread violation of privacy settings in the Twitter social network. In: Web 2.0 Security and Privacy (W2SP 2011), Oakland, CA, USA (2011) 6. Aharari, A., Inada, M., Abe, J.M., Nakamatsu, K.: Develop an embedded IoT system and its applications. In: IEEE International Conference on Innovations in Intelligent SysTems and Applications (INISTA 2018), Thessaloniki, Greece, pp. 1–5 (2018) 7. Ahrary, A., Inada, M., Yamashita, Y.: An IoA cloud-based farmer support system “AgriMieru”. In: Intelligent Decision Technologies Smart Innovation, Systems and Technologies, vol. 57, Tenerife, Spain, pp. 217–225 (2016) 8. Ahrary, A., Dennis, A.L.R.: A cloud-based vegetable production and distribution system. In: Intelligent Decision Technologies Smart Innovation, Systems and Technologies, vol. 39, Sorrento, Italy, pp. 11–20, June (2015)

Chapter 2

A Purpose of a Smart Vegetable Garden Model Based on Paraconsistent Annotated Evidential Logic Eτ Jonatas Santos de Souza, Jair Minoro Abe, Luiz Antônio de Lima, and Kazumi Nakamatsu Abstract Due to urban growth, the spaces for planting were being reduced; this growth caused an increase in the consumption of water, energy, and food. Consequently, it will be necessary to rethink a new structure for the use of urban space concerning planting. The information technology has converted the agricultural sector presently profoundly. Many studies are being carried out to facilitate the management and increase the productivity of the crops; the term currently used to name the phenomenon of technological implementation in the field is known as “Precision Agriculture.” This work aims to propose an intelligent garden model that allows the collection of information by the soil sensors that will be sent to the Arduino that controls a hydraulic pump responsible for irrigating the soil monitored by the sensor based on Paraconsistent Logic.

2.1 Introduction With the advancement of technology emerges the “fourth industrial revolution” or as it is called “Industry 4.0,” but in agriculture it is known as “Agriculture 4.0,” which uses some concepts of cyber-physical systems, cloud computing, and Internet of Things (IoT) [8, 9].

J. S. de Souza · J. M. Abe (B) · L. A. de Lima Paulista University, São Paulo, Brazil e-mail: [email protected] J. S. de Souza e-mail: [email protected] L. A. de Lima e-mail: [email protected] K. Nakamatsu University of Hygo, Himeji, Japan e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_2

11

12

J. S. de Souza et al.

That makes use of electronic devices and appliances, to further automate production. The use of information technology in agriculture is called Precision Farming. In 2012, the Ministry of Agriculture, Livestock, and Supply (MAPA), when instituting the Brazilian Commission of Precision Agriculture (CBAP), defined the Precision Agriculture (AP) as “an agricultural management system based on spatial and temporal variation of the productive unit and seeks to increase economic return, sustainability, and minimization of the effect to the environment” [5]. One of the main requirements of precision farming is the mapping of spatial and temporal variability in production units. This variability gives rise to the emergence of the Global Positioning System known as GPS [7].

2.2 Smart Garden A vegetable garden is a place where the typical plants are cultivated, such as vegetables, medicinal herbs, and spices. An Intelligent Garden combines the millennial art of cultivation and technology to obtain food in an automated way. With the automated irrigation system, it is possible to have a substantial reduction in water use. The Smart Garden is ideal for those who have little space at home or apartment and want to have a garden at home or want to have an understanding of IoT in practice and low cost. It is essential to know what will be cultivated, how to prepare the soil for what is planted, and the type of irrigation system so that the plant has better absorption of the nutrients from the water. Through sensors, it will give input data and show information so that the user understands the status of the garden or even control the garden system through software or mobile application. The Garden becomes clever because it has the decision to water the plants according to the output of the data.

2.3 Paraconsistent Logic The term “Paraconsistent” meaning “beyond the consistent” was coined in 1976 by Peruvian philosopher Francisco Miró Quesada. The Paraconsistent Logic is included among the non-classical logical calls, by defeating some of the fundamental principles of classical Logic, such as the principle of contradiction: according to the Paraconsistent Logic, a sentence and its negation can be both True. In the mid-1950, the Polish logician S. Jaskowski and the Brazilian logician N.C.A. da Costa proposed the contradiction in the logical structure and became known as the founders of the Paraconsistent Logic [2, 4].

2 A Purpose of a Smart Vegetable Garden Model …

13

2.4 Paraconsistent Annotated Evidential Logic Eτ The Paraconsistent Annotated Evidential Logic Eτ (Logic Eτ) [1] is a class of Paraconsistent Logic that works with propositions of type P(μ, λ), where p is a proposition and (μ, λ) indicate the degrees of favorable evidence and contrary evidence, respectively. The pair (μ, λ) is called the annotation constant, with the values of μ and λ ranging [3] between 0 and 1. One way to represent the Paraconsistent Logic that allows perceiving the real reach thus extracting results to subsidize decisionmaking is found in the understanding of the diagram and its degrees of certainty and uncertainty, grouped in extreme states identified in the results and non-extreme states shown in the results (Fig. 2.1). Table 2.1 describes the lattice symbols; they are indicated as the extreme and non-extreme states. The Extreme states can support the irrigation system in the case of information collected in the respective sensors and analyzed to release the water flow. In the use of the Paraconsistent Logic, the result can be parameterized for each state: • State “V” can be used to release water; • State “F” can be used to cut water flow; • State “T” can choose to cut water or release a room according to temperature and humidity; • State “⊥” can be parameterized as cutting the flow of water until new readings of the sensors. Fig. 2.1 Lattice with degrees of certainty and uncertainty, with adjustable limit control values, indicated on the axis [3]

14 Table 2.1 Extreme states and non-extreme states

J. S. de Souza et al. Extreme states

Symbol

True

V

False

F

Inconsistent

T

Paracomplete

⊥

Non-extreme States

Symbol

Quasi-true tending to inconsistent

QV → T

Quasi-true tending to paracomplete

QV → ⊥

Quasi-false tending to inconsistent

QF → T

Quasi-false tending to paracomplete

QF → ⊥

Quasi-inconsistent tending to True

QT → V

Quasi-inconsistent tending to False

QT → F

Quasi-paracomplete tending to True

Q⊥ → V

Quasi-paracomplete tending to False

Q⊥ → F

2.5 Algorithm Para-analyzer The para-analyzer algorithm translates the paraconsistent analysis by examining the values of favorable evidence grades, opposing evidence, resulting in possible values calculations using degrees of contradiction and certainty [3] (Fig. 2.2). */Values Settings*/ VUCE = C1/*Upper limit control of certainty*/

Fig. 2.2 Representation of a typical paraconsistent system. Adapted from [3, 6]

2 A Purpose of a Smart Vegetable Garden Model …

VLCE = C2/*Lower limit control of certainty*/ VUUN = C3/*Upper limit control of uncertainty*/ VLUN = C4/*Lower limit control of uncertainty*/ */Input Variables*/ μ/*Favorable Evidence*/ λ/*Unfavorable Evidence*/ */Output Variables*/ S1 */Discrete Output*/ S2 */Analog Output*/ S3 */Analog Output*/ */Mathematical Expressions*/ Thus: 0 ≤ μ ≤ 1 and 0 ≤ λ ≤ 1 DUN = (μ + λ) − 1 DCE = μ − λ */Determination of Extreme Logical data*/ If DCE ≥ C1 then S1 = V If DCE ≤ C2 then S1 = F If DUN ≥ C3 then S1 = T If DUN ≤ C4 then S1 = ⊥ */Determination of Non-Extreme Logical data*/ For 0 ≤ DCE < C1 and 0 ≤ DUN < C3 If DCE ≥ DUN then S1 = QV → T else S1 = T → V For 0 DCE < C1 and C4 < DUN ≤ 0 If DCE ≥ |DUN | then S1 = QV → ⊥ else S1 = ⊥ → V For C2 < DCE ≤ 0 and C4 < DUN ≤ 0 If |DCE ≥ ||DUN | then S1 = QF→ ⊥ else S1 = ⊥ → f For C2 < DCE ≤ 0 and 0 ≤ DUN < C3

15

16

J. S. de Souza et al.

If |DCE | ≥ DUN then S1 = QF → T else S1 = T → F DUN = S2 DCE = S3 */End*/

2.6 Proposing the Model It is essential to define the type of irrigation system to be used and the type of vegetables to be planted. In Fig. 2.3, lettuce irrigation is proposed. Through the display connected to an Arduino platform, basic information about the irrigation system, the amount of water irrigated, and the time of the next irrigation is presented. Paraconsistent Logic assisted in decision-making to drive the irrigation system. The inputs will be picked up by sensors that measure the soil moisture and sensors that will measure the temperature; the inputs are analyzed and processed. After processing, it will perform the normalization step of the data, which is the transformation of the data captured by the sensors in a range of 0 and 1. According to the concept of Logic Eτ [1], the output after the normalization determined whether the planting will be watered or not, according to the 12 states that are presented in Fig. 2.1. Given an exit of the possibility of irrigating the garden, the irrigation system will be triggered, and the water will be flowing through the PVC pipe to the plant. For the plant to better absorb the nutrients from the water, it will be used in the proposed model, the irrigation system by drip microaspiration; the system will only stop when the output informs no longer to irrigate. By this way, the application will make possible to see the temperature of the garden, soil moisture, the status of the garden, and the option to control the irrigation system manually. The proposed project seeks to adapt to different types of vegetable gardens, such as the traditional garden, suspended, organic, and domestic, and also the types of vegetables to be grown.

2.7 Conclusion The viability of the technological use of non-classical logics is perceived, in this case, the paraconsistent logic Eτ, as a support in the irrigation of the water flow, since the data collected by the sensors can be manipulated organically by such logic and that allow the application of computational mathematical models for decision making. The logic Eτ provides an attractive model based on Design Thinking.

2 A Purpose of a Smart Vegetable Garden Model …

17

Fig. 2.3 Design thinking of the proposed model (Author)

Acknowledgments This study was partially supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES)—Grant Code 001, Process No. 23038.013648/201851.

References 1. Abe, J.M.: Paraconsistent logics and applications. In: Proceedings of the 4th International Workshop on Soft Computing Applications, Budapest, Arad, vol. 18, pp. 11–18. IEEE (2010) 2. Abe, J.M. (ed.): Paraconsistent Intelligent Based-Systems: New Trends in the Applications of Paraconsistency, Germany. Intelligent Systems Reference Library, vol. 94, 309 p. Springer. ISBN 978-3-319-19721-0 3. Abe, J.M.: Introduction to Annotated Logics—Foundations for Paracomplete and Paraconsistent Reasoning. Springer International Publishing, São Paulo (2015) 4. Akama, S.: Towards Paraconsistent Engineering, Intelligent Systems Reference Library, vol. 110, 234 p. Springer International Publishing, Switzerland (2016). ISBN 978-3-319-40417-2 (Print), 978-3-319-40418-9 (Online), Series ISSN 1868-4394. https://link.springer.com/book/ 10.1007%2F978-3-319-40418-9 5. Brasil: Ministério da Agricultura, Pecuária e Abastecimento. Agricultura de precisão/Ministério da Agricultura, Pecuária e Abastecimento. Secretaria de Desenvolvimento Agropecuário e Cooperativismo, 36 p. Mapa/ACS, Brasília (2013). ISBN 978-85-99851-90-6 6. Lima, L.A.: Sistema especialista AITOD baseado na Lógica Paraconsistente Anotada Evidencial Eτ (in Portuguese). MSc Dissertation, Paulista University, São Paulo (2018) 7. Ray, P.P.: Internet of things for smart agriculture: technologies, practices and future direction. J. Amb. Intell. Smart Environ. 9(4), 395–420 (2017)

18

J. S. de Souza et al.

8. Stafford, J.V.: Implementing precision agriculture in the 21st century. J. Agric. Eng. Res. 76, 267–275 (2000). https://doi.org/10.1006/jaer.2000.0577 9. Zheng, L.: Technologies, applications, and governance in the internet of things. In: IoT Global Technological and Societal Trends (2011)

Chapter 3

Paraconsistent Artificial Neural Network Applied to Agribusiness Taciana Tamyris Alves de Souza, Cristina Corrêa de Oliveira, Jair Minoro Abe, Ari Aharari, and Kazumi Nakamatsu

Abstract Brazil is the world’s second largest exporter of soybeans, but growers face challenges, individually crop pests. Information Technology can provide solutions that contribute to the fight against brown stink bugs since this is the insect that causes the most loss of Brazilian production. The Paraconsistent Logic and the Paraconsistent Artificial Neural Networks allow us to create solutions for the identification of the insect in the plants, from in loco images. This article explored the techniques available for the construction of applications in Agriculture 4.0 for insect recognition, in order to facilitate the identification of them in the plantations, through the recognition of patterns of the aerial parts of the plant. The objective of this research is exploratory, with a qualitative approach and bibliographic and documentary procedures. Sources were used as scientific articles, research on international institutional sites, and research of data in national governmental institutions that work with the production, control, and export of this commodity. The biggest challenge of Agriculture 4.0 is to integrate technologies that can contribute to agribusiness. This is because farmers face difficulties accessing the internet in the field, which harms the use of devices in the plantations. This problem also affects countries in Europe, such as Germany, that need to ensure fiber-optic internet in rural areas. Despite the T. T. A. de Souza Fatec Zona Leste, São Paulo, Brazil e-mail: [email protected] C. C. de Oliveira Federal Institute of São Paulo, São Paulo, Brazil e-mail: [email protected] J. M. Abe (B) Paulista University, São Paulo, Brazil e-mail: [email protected] A. Aharari Sojo University, Kumamoto, Japan e-mail: [email protected] K. Nakamatsu University of Hyogo, Kobe, Japan e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_3

19

20

T. T. A. de Souza et al.

challenges encountered, solutions are using Artificial Intelligence in the area of agriculture that has contributed to the Brazilian agribusiness but is still an unexplored strand.

3.1 Introduction The information revolution has been gaining a broader scope in the last years, allowing several benefits for the current society constitution. Nowadays, there is the possibility of transforming any data into a useful source of solving problems, setting up a great Data Science. National agriculture expanded its frontier between 1960 and 1990, called the Green Revolution, making Brazil one of the largest exporters of commodities. In this period, the character of the agribusiness was the agricultural mechanization, monoculture with intense land preparation, and application of inputs like fertilizers and pesticides. The second wave, known as Integrated Systems, started in 1990, is remaining active until the present moment. It differs from the previous wave using integrated and rotated production systems, intensifying the cultivation of soy integrated into animal production. A striking feature of this wave is to allow environmental balancing, optimizing the use of natural resources. It is a phase that requires disciplinary vision in the generation of new knowledge and technological solutions. In 2012, the term Industry 4.0 was adopted in Germany, whose concept brought innovation throughout the German automotive industry, expanding to factories of different segments and applying in the automation in production processes [1]. The methods of Industry 4.0 have been applied in agribusiness with the usage of technologies such as the Internet of Thing (IoT), Cyber-Physical Systems, Cloud Computing, and the Artificial Intelligence (AI) in the construction of computational solutions to apply logical rules, learning, and pattern recognition [2], which resulted in the term Agriculture 4.0 since agricultural production is directly related to agroindustry production and the final consumer, guaranteeing the use of the technology subsidizing the production with the maximum possible yield [3]. The application of technology in agriculture has intensified in the last four years since the search of the terms “Smart Farm” and “Agriculture 4.0” in the base’s Web of Science and Science Direct demonstrated and an increase of the articles related to technology in its bases, as can be seen in Fig. 3.1. Soy is the country’s leading agricultural crop, contributing directly to agribusiness and Gross Domestic Product (GDP). The forecast in 2017 was that the GDP of agribusiness would grow around 0.5–1% and that this crop would correspond to 89% of the grains produced in the country, with a range between 106.4% and 3 million t [4]. The largest buyer of Brazilian soybeans is China, accounting for 76% of Brazil’s soybean export revenue. This study aims to analyze the problems related to pests in soybean cultivation and the techniques available for the construction of future applications using IoT, to help

3 Paraconsistent Artificial Neural Network Applied to Agribusiness

21

Fig. 3.1 Scientific articles on agriculture and technology

identify the insects present in the plantations, through the recognition of patterns in the aerial parts of the plant, in recognition of the presence of the brown stink bug. The primary objective is the identification of the algorithms applied in Agriculture 4.0, specifically in Smart Farm, allowing to verify its evolution, by compiling recent algorithms, providing subsidies for future studies in this field, as well as to propose the Application Paraconsistent Annotated Evidential Logic Eτ, specifically the Paraconsistent Artificial Neural Network in this area.

3.2 Soybean Soy is the most present grain in the Brazilian economy, growing every day, and conquering not only the Brazilian market but also European consumers. In 2017, soybeans led exports for the third consecutive year, with variations in sales revenue abroad and average prices compared to the previous [5]. The forecast for the 2018/2019 harvests is still part of expectations, as it depends on weather conditions throughout the soybean crop cycle [6]. The forecast will be approximately 62 million ha, with soy and corn occupying more than 80% of this area [4]. As a result, the area planted with soybeans will exceed 36 million ha, about 1.0 million ha larger than the area of the previous cycle (35.15 m ha), foreseeing a production of 3 million t higher than the current crop yield [7]. Insects are a significant problem for soybeans, often leading to significant losses in production. The productive sector related to the soybean crop is attacked by dozens of pests every year. Many insects can wipe out an entire crop of soybeans if they are not controlled. However, identifying the species that is causing damages in the plantation is not trivial, as the pests attack in different ways and a myriad of species. For example, soybean pests can reside both underground, directly attacking the roots, and through the air, attacking the main stem, leaves and secondary stems. The most common disease is mad soybeans, which causes the leaves to become rough as well as thickens their veins, as demonstrated in Fig. 3.2, causing significant damage to the growers [8]. The brown stink bug, shown in Fig. 3.3, is one of the most abundant in the soybean crop, requiring more attention. Therefore, it is necessary to identify these pests in

22

T. T. A. de Souza et al.

Fig. 3.2 Leaf injury

Fig. 3.3 Brown stink bug

the soybean in its various stages and to understand a little better the challenges faced by the cultivators of this grain [9].

3.3 Methodology This paper proposes to carry out scientific research of an applied nature, with exploratory objectives that are carried out to clarify ambiguous situations or discover business opportunities with a qualitative approach. Finally, the research process is bibliographical and documentary. The bibliographic research, called the secondary source, was carried out based on the collection of academic references published by written and electronic means, such as books, websites, and electronic journals. Surveys were carried out regarding soybean production, extending to agribusiness, using bibliographic procedures with scientific research, and documentary research with data from governmental institutions that work with production, research on international institutional sites, control, and export of this commodity.

3 Paraconsistent Artificial Neural Network Applied to Agribusiness

23

The researches related to the algorithms were carried out to verify the development and progress of Agriculture 4.0 with a focus on the control of soybean pests. A period of 10 years for sample studies was defined in order to incorporate only recent studies. The research was developed within the context of agriculture so that it was possible to find relevant applications already existing in the field of study, where the first step was to identify articles based on a search of search strings, “Brown Stink Bug” and “IoT,” “Brown Stink Bug” and “Smart Farm,” in all fields of research. Relevant electronic databases such as AGRIS (United Nations, Food and Agriculture Organization), Engineering Research Database, Public Library of Science (CrossRef), SciELO (CrossRef), ScienceDirect Journals (Elsevier), Scopus (Elsevier), SpringerLink, Technology Research Database, and Wiley Online.

3.4 Paraconsistent Artificial Neural Network The core formulas of the Paraconsistent Annotated Evidential Logic (Logic Eτ) are of the type p(μ, λ), where (μ, λ) ∈ [0, 1] and [0, 1] are the real unitary intervals (p designates a propositional variable). P(μ, λ) can be understood as μ represents the favorable evidence and λ represents the contrary evidence for the proposition p. The Favorable Evidence Degree is a value between 0 and 1 that represents the favorable evidence in which the sentence is true. The Contrary Evidence Degree is a value between 0 and 1 that represents the contrary evidence in which the sentence is untrue. Through the Uncertainty degree: Gun(μ, λ) = μ + λ − 1 (0 ≤ μ, λ ≤ 1) and Certainty degree: Gce(μ, λ) = μ − λ (0 ≤ μ, λ ≤ 1), it is possible to characterize the four extreme logical states False, True, Inconsistent, and Paracomplete as well as the 12 non-extreme logical states remaining. All the states are represented in Fig. 3.4.

Fig. 3.4 Extreme and non-extreme states [10]

24 Nevus

T. T. A. de Souza et al. Melanoma

Melanoma

Binary image

Fig. 3.5 Nevus and malignant melanomas [13]

The operator Max indicates the operation of maximizing real numbers with the usual order. Given two values from different sources (μ1 , λ1 ) and (μ2 , λ2 ) ∈ τ, (μ1 , λ1 ) OR (μ2 , λ2 ) = (Max{μ1 , μ2 }, Min{λ1 , λ2 }). The operator Min indicates the operation of the minimization of real numbers with the usual order. Given two values from different sources (μ1 , λ1 ) and (μ2 , λ2 ) ∈ τ, (μ1 , λ1 ) AND (μ2 , λ2 ) = (Min{μ1 , μ2 }, Max{λ1 , λ2 }). A Paraconsistent Artificial Neural Network (PANN) is a connectionist-like structure composed of a family of Paraconsistent Artificial Neural Cells based on Logic Eτ. The primary objective of the paraconsistent analysis is to know by what measure, or degree of certainty, that a proposition is False or True. In this analysis, only the value of the degree of Gce is considered as a result. The value of the degree of uncertainty Gin is indicative that informs the measure of the inconsistency. If there is a low value of certainty or much inconsistency, the result is a lack of definition [11]. The PANNs have been used in different areas of knowledge, from the health area with the diagnosis of skin and breast cancer, cephalometric analysis, and recognition of writing [12]. The melanoma identification research used the PANN to do image recognition and classification. The process begins with the mapping of the image adequately treated, transformed into a binary image with little noise, as can be seen in the fourth image of Fig. 3.5. The system performs the mapping of the edges of the image since this data is relevant to differentiate nevus from melanomas. The result of this mapping is the degree of favorable evidence for black pixels, generating a list by edges, with the image divided into four parts. These data are organized in a histogram, which will be compared with images one to three of Fig. 3.5. The first image represents a nevus, and the other two represent the lesions on the skin with melanoma.

3.5 Applications in Agribusiness Considering the agribusiness scenario, Fig. 3.6 shows the leaf images and their transformation into digital images used for the diagnosis of plant diseases. Digital images of coffee, sugar cane, soybeans, and fruit plants have been transferred to a database of images that are stored and processed so that it becomes a support for decision-making

3 Paraconsistent Artificial Neural Network Applied to Agribusiness

25

Fig. 3.6 Leaf diseases [15]

of proper control. Thus, it is possible to structure a database containing images of cultures, diseases, and conditions found in the field and to construct tools capable of assisting in the identification of diseases with similar symptoms through expert systems [14]. The correlation of these scenarios demonstrates the use of Information Technology (IT) to solve the problem of the target public, allowing not only the integration of data-based decision systems but also the increase of assertiveness and efficiency in the processes performed. Another approach in the context of agribusiness is the application of machine learning for the recognition of coffee berries in field images [16], using about 3393 manual images labeled as coffee and non-coffee, which enabled the realization of quantitative tests, and the detection of the coffee berries had an accuracy of 90% employing Support Vector Machines and Oriented Digest Histograms. The searches carried out in the databases showed that the applications for the recognition of the insect using IoT, or even the concept of Smart Farm, are incipient. One study used the application of substrate-based vibrational signals intraspecific to cotton planting [17]. Based on the examples of the applications presented, it is possible to note the importance of the technologies used, demonstrating that IT and Artificial Intelligence can solve complex problems with greater assertiveness through the Artificial Neural Network Paraconsistent.

3.6 Results and Discussion According to the Brazilian Institute of Geography and Statistics (IBGE), only the agricultural sector accounts for approximately 22.5% of GDP and 37% of the labor force. Considering that for at least 50 years, Brazil was one of the major importers of food, producing in a considerably smaller quantity than today, and due to the advances in technological solutions applied along this route, it is possible to observe the financial results at present. The automation of agricultural processes has become a promising trend due to the evolution of technologies such as agro-informatics, bioinformatics, and precision agriculture. Technology has influenced strongly to contribute to the evolution of agriculture. The technology used in the agricultural field has consolidated the new era within the field scenario for process optimization. Its great potential resides in its

26

T. T. A. de Souza et al.

transversality, being able to add value and benefit to the different areas of business, market, agriculture, and the environment. The biggest challenge of Agriculture 4.0 is to research so that all these technologies are integrated and generate knowledge for Brazil to continue as one of the leading exporters and producers of agriculture. For this to be done, it is necessary to take into consideration that, despite the opportunities, there are many difficulties to access the internet in rural areas, limiting the advancement of applications for use in the field. Agriculture 4.0 is still a significant challenge because access to the Internet, according to census agricultural [18], shows 1.42 million producers accessing the Internet via broadband (n = 46.2%) and mobile internet (n = 63.77%), but the 4G signal does not cover the whole territory, making it challenging to use IoT applications. Countries in Europe also face these problems. Germany is being charged for providing fiber-optic internet in a rural area with innovative and cost-effective techniques for data infrastructure facilities. Another demand from producers is the financial incentive for agribusiness; ensure that farmers have sovereignty over their data [19]. Considering the advancement of Agriculture 4.0 and its technologies’ approaches used in this era, there are currently some systems for the identification of insects in plantations. The biggest problem for grain producers is to identify the pests that attack the crops so that a control method is applied since they can be confused with the natural enemies of these insects. In order to identify the insect species, it was possible to compare the images taken in the field with the ones in the application. The primary objective of this application is to display only the image corresponding to the catalog, to help the farmer to distinguish between the pests that are beneficial to his plantations with the natural enemies since there is no use of beneficial insects in the crop if the farmer confuses them with those that can cause damage to the plantation. It can be incorporated into the problem of soybean in future works since the identification of the brown stink bug would occur through in loco photographs of the insect in the plantations. Also, the damages caused by it in the leaves and roots and from the photographed image, it would be possible to recognize the species and stage of the pest. Thus, it is possible to facilitate soybean management and mitigate losses for soybean farmers since the brown stink bug implies the loss of yield and quality of the soybean seed. At present, no solutions are using the technology to treat the soybean problem in an efficient way so that the same procedure involving brown stink bug could be applied to the other soybean pests, optimizing and standardizing the harvest of this valuable grain for the economy. This study is concerned with identifying alternatives to assist soybean farmers in the control of brown stink bug, using Artificial Intelligence to detect the appearance of this insect in soybean plantations in order to apply specific pesticides to the pest. Firstly, a survey was made on the importance of soybeans for the Brazilian and world economy as well as the reason for the research to be based on this grain. It also analyzed the existence of other technological solutions applied in Agriculture 4.0 that are used in the optimization of processes.

3 Paraconsistent Artificial Neural Network Applied to Agribusiness

27

About the existing technologies, it is possible to notice that despite investments already existing in Agriculture 4.0, there is still a long way to go, especially in the soybean plantation, which provides net results that are very advantageous for the economy, enabling a return on investment in the medium term. Regardless of the crop that occurred in the year, the soybean crop shows results above average and is reliable for future investments. Through the analysis of existing solutions for the field of agriculture, it was possible to perceive that the alternatives for the use of AI within agriculture are still carried out generically, considering not only several insects but also other grain crops. There must be specificity for the different approaches so that it is possible to treat soybean problems uniquely. In this way, using the identification of the brown stink bug in the plantations, it would be possible to apply specific procedures to combat this pest.

References 1. VDMA-Verlag: Industrie 4.0 konkret—Lösungen für die industrielle Praxis, vol. 59, p. 5. VDMA-Verlag (2018) 2. Bonneau, V., Copigneaux, B., Probst, L., Pedersen, B.: Industry 4.0 in agriculture: focus on IoT aspects (2017) 3. Braun, A.-T., Colangelo, E., Steckel, T.: Farming in the era of Industrie 4.0. Procedia CIRP 72, 979–984 (2018) 4. Política, Agrícola, S.: Brasil: Projeções do Agronegócio: Brasil 2017/18 a 2027/28 projeções de longo prazo, 1st ed. MAPA, Brasília (2018) 5. Associação Comércio Exterior do Brasil: Revisão da Balança Comercial Para 2018 (2018) 6. Brasil: Ministério da Agricultura, Pecuária e Abastecimento. Projeções do Agronegócio: Brasil 2017/18 a 2027/28 projeções de longo prazo / Ministério da Agricultura, Pecuária e Abastecimento. Secretaria de Política Agrícola. – Brasília: MAPA/ACE (2018) 7. Brasil: Acompanhamento da safra brasileira, vol. 6. Companhia Nacional de Abastecimento, Brasília (2018) 8. Landgraf, L.: Soja Louca II é reconhecida como doença da soja pelo Mapa, Londrina (2015) 9. Peixoto, M.F.: Percevejo-marrom: Um perigo real para os grãos de soja (2017) 10. Abe, J.M., Akama, S., Nakamatsu K.: Introduction to Annotated Logics—Foundations for Paracomplete and Paraconsistent Reasoning, vol. 88, 1st edn. Springer International Publishing, Switzerland (2015) 11. da Silva Filho, J.I., Abe, J.M., Torres, G.L.: Inteligência Artificial com as Rede de Análises Paraconsistentes. LTC, Rio de Janeiro (2008) 12. Abe, J.M. (ed.): Paraconsistent Intelligent-Based Systems—New Trends in the Applications of Paraconsistency, vol. 94, 1st ed., Springer International Publishing, Switzerland (2015) 13. Souza, S., Abe, J.M.: Nevus and melanoma paraconsistent classification. Stud. Health Technol. Inform. 207, 244–250 (2014) 14. Barbedo, J.G.A., Gomes, C.C.G., Cardoso, F.F., Domingues, R., Ramos, J.V., McManus, C.M.: The use of infrared images to detect ticks in cattle and proposal of an algorithm for quantifying the infestation. Vet. Parasitol. 235, 106–112 (2017) 15. Barbedo, J.G.A.: Using digital image processing for counting whiteflies on soybean leaves. J. Asia-Pacific Entomol. 17, 685–694 (2014) 16. Santos, T.: Detecção automática de bagas de café em imagens de campo. em Congresso Brasileito de Agroinformática, Ponta Grossa (2015)

28

T. T. A. de Souza et al.

17. Lampson, B.D., Han, Y.J., Khalilian, A., Greene, J., Mankin, R.W., Foreman, E.G.: Automatic detection and identification of brown stink bug, Euschistus servus, and southern green stink bug, Nezara viridula (Heteroptera: Pentatomidae) using intraspecific substrate-borne vibrational signals. Comput. Electron. Agricult. 91, 154–159 (2013) 18. IBGE- Instituto Brasileiro de Geografia e Estatística. Censo Agropecuário 2017: resultados preliminares. Brasília: IBGE (2017) 19. Bauernverband, D.: Landwirtschaft 4.0 – Chancen und Handlungsbedarf (2016)

Part II

Regular Papers

Chapter 4

Research and Design on an Automatic Control System Using in Antarctic Greenhouse Kaiyan Lin, Chang Liu, Jie Chen, Junhui Wu, and Huiping Si

Abstract To solve the difficulty of Chinese scientific examiner eating fresh vegetables in Antarctica station, a greenhouse in the research station was built. It is the first natural light greenhouse in Antarctic land. Running in the severe environment, the greenhouse has the problem of unattended operation and no professional cultivation, and thus demands high reliability. An automatic control system for the Antarctic greenhouse was developed to create a good environment for crop growth. The system is composed of a computer, an environment controller, a data collector, sensors, actuators, and so on. In the greenhouse, environmental parameters are collected and sent to computer for processing and displaying by an environment controller connecting to the data collector for acquisition. In extremely cold area, temperature is the primary factor for a greenhouse, which is heated by water pipes. The fuzzy logic was used to automatically adjust the heating pipe opening according to the measured ambient temperature for temperature control. Sodium lamp and LED were used to supplement light in greenhouse according to Photosynthetic effective value. With controlling the greenhouse’s top cover, side cover, and the natural light supplement, it could simulate the diurnal variation and the seasons change to facilitate plant growth. The greenhouse is irrigated in the tidal irrigation way. For different plants, it is convenient to adjust the irrigation time and duration for greenhouse workers. According to the control strategy, the main factors such as temperature, illumination, and fertilizer of the greenhouse environment are automatically adjusted. Since it was installed in the Antarctica Great Wall Station, the system has proved to run stably with long time monitoring data, providing a guarantee for fresh vegetables supply for resident scientific examiner.

K. Lin (B) · C. Liu (B) · J. Chen · J. Wu · H. Si Modern Agricultural Science and Engineering Institute, Tongji University, Shanghai, China e-mail: [email protected] C. Liu e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_4

31

32

K. Lin et al.

4.1 Introduction Polar scientific investigation is an important part of China’s marine development strategy. Due to the cover of ice and snow, as well as frequent snowstorms, the natural environment that in most Antarctic areas is severely poor; hence, the provision of fresh vegetables for scientific expeditions has always been a problem. Currently, several Antarctic scientific research stations, such as Davis Station of Australia, McMurdo Station of the United States, and Showa Station in Japan, utilize a container-like plant factory to realize partial production of vegetables with artificial light. Supported by the Ministry of Science and Technology of China, the research team designed and built a greenhouse in the Antarctic Great Wall Station. The greenhouse of Great Wall Station is the first natural light greenhouse in the Antarctic continent, aiming at automatically creating a controlled environment suitable for crop production. Several problems faced by the automatic control system of the greenhouse are described as following. Firstly, the greenhouse is in an extremely cold climate. Secondly, there are polar day and polar night, thus this automatic control system of the greenhouse requires reasonable light, temperature, water, and fertilizer regulation to create the best growing environment for plant growth. Thirdly, the network infrastructure is weak. Not only does the entire Great Wall Station have a bandwidth of only 2 M, but also only tens of K are allocated to the greenhouse. Lastly, the greenhouse is operated by an emergency physician with no professional maintenance. Therefore, the system is required to have high reliability and stability, and then should be able to perform all-weather automatic control. The crop’s growth process is complex and dynamic, affected by external factors such as environment, nutrients, and water [1]. The sugar required for crop’s growth is produced by photosynthesis and affected by solar radiation, temperature, water and fertilizer, CO2 , and other factors. Controlled and optimized crop’s growth environments must fully consider these factors. Among them, temperature is an important one in the regulation of greenhouse environment, directly affecting crop photosynthesis, transpiration rate, crop yield, etc. And thus, temperature must be controlled in a suitable range because too high or too low temperature will affect severely the yield and quality of the crops, and even leading to disease. Temperature control in greenhouse is a dynamic, nonlinear, and time-varying system, which makes it difficult to design an accurate mathematical model [2–6]. When the system is obscure or the knowledge is deficient, using fuzzy logic to deal with such problems will have better scalability. Thus, fuzzy logic has been applied successfully in greenhouse temperature control [7–9]. In terms of membership choice, Revathi [10] confirmed that the optimized membership function is better than the normal fuzzy model application in the same system. In addition to temperature, light, water, and fertilizer act as environmental signals on crops, which also affect crop growth and development. Tang et al. [11] proposed an intelligent auxiliary lighting system for greenhouse plants. The system was designed for greenhouse plants with 3D auxiliary lighting that can automatically switch to sunlight or LED under various weather conditions, which improved the growth rate of fruits and vegetables. As regard to irrigation, Ma

4 Research and Design on an Automatic Control System …

33

et al. [12] studied the effects of tidal irrigation and drip irrigation on the moisture content of the substrate, which proved that tidal irrigation can significantly improve the utilization efficiency of water and fertilizer resources and then reduce the leakage loss of water and fertilizer. The greenhouse is ideal for crop growth and is a closed environment that controls climate and irrigation variables. Climate and irrigation are two separate and mutually influential systems with different control issues and objectives [13–15]. For the natural light greenhouse control in the Antarctic environment, less experience can be learned. In this paper, based on the requirements of the automatic control of the Antarctic greenhouse, the control scheme was designed independently. After collecting the environmental information in the monitoring greenhouse, the fuzzy logic was used for temperature adjustment. Also, the lighting control was realized with the top cover, side cover, and fill light. Meanwhile, the tidal irrigation was used to benefit greenhouse crop production and wastewater treatment.

4.2 Designing Architecture of Greenhouse Automatic Control System The area of Great Wall Station’s greenhouse is about 36 meters and is divided into two areas for planting leafy vegetables and fruit vegetables, respectively. As shown in Fig. 4.1, each one was equipped with sensors measuring such parameters as total radiometer, photosynthetically active radiation, CO2 , temperature, and humidity for collecting environmental information. The nutrient tanks of vegetables Fig. 4.1 Profile of greenhouse structure

Top cover

Top cover

Total Total Photosynthetic Photosynthetic radiation radiation sodium lamp Side cover Air heating pipe

Ground heating pipe

sodium lamp Side cover

34

K. Lin et al.

were equipped with electric conductivity (EC) meter and PH meter. In the greenhouse, there are two sets of heating pipes, called air heating and ground heating, respectively. The air heating pipes were erected in the middle of the greenhouse, and arranged along the wall of about 1.8 m high from the bottom plate. The ground heating pipe was laid above the floor. The pipe was equipped with a water temperature sensor and a four-way valve for controlling the water flowing rate. All sensors were connected to an indoor environmental data collector, which was connected to the computer via an environmental controller. Due to the polar day and polar night in the Antarctic, black curtains were placed on the top and sides in order to create a suitable lighting effect, which were called top cover and side cover, respectively. The computer controls the expansion and contraction of the top and side covers, as well as the opening of the fill light, to simulate different illumination in daylight shifting and season changing to facilitate plant growth. As shown in Fig. 4.2, the embedded environment controller is connected to an environmental data collector with RS485 bus to sample environmental information in the greenhouse. The information is sent to a computer for processing after local storage, based on which an environmental control strategy is made and sent to the environmental controller. The environmental controller is also connected to a switch signal output board to control the corresponding actuators. Expansion and contraction signals of the top and side covers are input to the environmental controller through a signal input board. At the same time, the network surveillance camera is linked to the computer with a router. On one hand, the computer could directly get and display the image in the greenhouse; on the other hand, the plant image was captured and stored periodically according to a time interval. Remote server (Data center) IP network Light sensor Temperature and humidity sensor CO2 sensor Water temperature sensor

Video Surveillance RJ45 Surveillance Router camera

IP network

Data collection Computer

RS232 Communication Switch RS485signal output port B Environmental board data collector Environment Switch al controller signal input board Communication RS485 port A

Nutrient fluid sensor

Fig. 4.2 System framework for greenhouse automatic control

Actuator Top cover and side cover signal

Actuator control

4 Research and Design on an Automatic Control System …

35

4.3 Control Scheme 4.3.1 Greenhouse Environmental Data Collection As mentioned above, many sensors were installed in the greenhouse to monitor the greenhouse as shown in Table 4.1. The computer collects environmental information by communicating with the environment controller, using the data packet as shown in Fig. 4.3. After receiving data packet from the controller, the computer parses measurement parameters according to a designed communication protocol and determines the name and sampling value of the sensor according to the channel number (address) in Table 4.1 Sensor information Sensor

Use

Quantity

Channel number

Unit

Range

Water temperature sensor

Air heating pipe inlet and outlet temperature

4

0–3

°C

0–100

Total radiometer

Solar radiation in leafy and fruity areas

2

7, 13

W/m2

0–2000

Photosynthetic efficiency

Photosynthetic efficiency in leafy and fruity areas

2

8, 14

umol/m2 /s

2–2000

Ambient temperature

Temperature in leafy and fruity areas

2

9, 15

°C

−40–60

Environment humidity

Humidity in leafy and fruity areas

2

10, 16

%

0–100%

PH meter

PH value in leafy and fruity areas

2

11, 18

Conductivity EC

EC value in leafy and fruity areas

2

17, 19

mS/cm

0–20

CO2

CO2 concentration in the greenhouse

1

12

ppm

0–5000

Floor temperature

Floor temperature

1

5

°C

0–100

Fig. 4.3 Package format of measured data

0–14

36

K. Lin et al.

Table 4.1. The measurement values are calculated according to different conversion formulas. For example, the total radiation R was computed as Formula (4.1). R=

Sa − 4 3000 × Se 16

(4.1)

where S a is a parameter with 4 bytes of floating point number converted by the BitConverter. ToSingle function provided by .Net FrameWork and Se represents the sensitivity of one sensor. Since the environmental parameters of the greenhouse varying slowly, it is necessary to eliminate large pulse interference from the measurement data for smoothing. In this paper, the median average filtering method was used for smoothing. That is, for continuous sampling data sets, and are the maximum and minimum value respectively. After removing the maximum and minimum values, the average value S m is calculated as the measured value according to Formula (4.2), in which the range of N is set from 6 to 10. N 

Sm =

i=1

si − smax − smin N −2

(4.2)

This method can suppress random interference and filter out obvious pulse noises.

4.3.2 Ambient Temperature Control Based on Fuzzy Logic Among the environmental control factors (temperature, humidity, light, carbon dioxide, etc.) of the greenhouse, temperature is the most important one, which should be satisfied firstly in the control process [16]. Because the greenhouse is located in Antarctica, the ambient temperature is very low. There are two sets (air and ground) of heating pipes in the greenhouse, as shown in Fig. 4.4. Water temperature sensors were installed at the inlet and outlet of the pipes for monitoring the water temperature. The heating pipe was connected with a pump providing power and a four-way valve determining the water flowing rate in water circulation. An ambient temperature sensor was installed, respectively, in the leafy and fruit-growing areas for obtaining temperature. After processing with fuzzy logic, the computer sends instructions to environmental controller to start or stop the water pump and regulate the opening of the four-way valve, thereby adjusting the ambient temperature in the greenhouse. For a fuzzy controller, it generally includes three steps: fuzzification, fuzzy reasoning, and sharpening. The fuzzy system is a rule-based system whose fuzzy rules are processed by an inference engine. The inference engine evaluates the rules and makes a conclusion, which can be regarded as interpolation between subsystems [17]. Fuzziness is the input interface of a fuzzy controller. Its function is to convert

4 Research and Design on an Automatic Control System …

37

Circulation direction Inlet water temperature sensor Pump

Four-way valve Outlet water temperature sensor

Air heating pipe Circulation direction Inlet water temperature sensor

Pump

Leaf vegetable area sensor Fruit vegetable area sensor

Four-way valve Outlet water temperature sensor

Floor heating pipe

Fig. 4.4 Greenhouse heating system

the precise input of the system into a fuzzy set and its corresponding membership degree in an appropriate proportion. Fuzzy reasoning applies fuzzy input value to the “IF-THEN” control rules base, and “accumulates” the results produced by each rule to form a fuzzy output set. Finally, the sharpening is to defuzzify these fuzzy outputs by finding the most representative output control variable that can directly drive the actuator within output range. For easy operation, within the error range of ± 3 °C for ambient temperature, the fuzzy control value is divided into seven fuzzy subsets, which are NB (negative big), NM (negative medium), NS (negative small), ZO (zero), PS (positive small), PM (positive medium), and PB (positive big). The ambient temperature deviation E and the difference change rate EC were selected as the input variables. The output variable is the heating amount of the pipe, corresponding to the opening degree U of the pipe. Correspondingly, the fuzzy sets are E, EC, and U. Among them, the domain of the ambient temperature deviation E is consistent with the temperature change rate EC. Because the triangle membership function has characteristics of simple calculation and easy implementation [18], it was selected as the membership function, as shown in Fig. 4.5a. The output control value also adopted the triangle membership function, as shown in Fig. 4.5b, with four fuzzy set of ZO (zero), PS (positive small), PM (positive medium), and PB (positive big), corresponding to the four interval openings of the four-way valve, respectively. By means of the on-site experiments and practical experience, the fuzzy reasoning rules for greenhouse environmental temperature control were formed as shown in Table 4.2. The fuzzy reasoning module used the max–min operator for rules inference, in which min and max method was used for fuzzy AND and fuzzy OR operator, respectively, and the sharpness method used center of gravity method.

38

K. Lin et al.

(a)MF of E(EC)

(b)MF of U

Fig. 4.5 Membership function (MF) of input and output variable

Table 4.2 Fuzzy inference rules E

NB

NM

NS

ZO

PS

PM

PB

EC NB

PB

PB

PB

PB

PB

PB

PB

NM

PB

PB

PB

PB

PB

PM

PS

NS

PB

PB

PM

PM

PS

PS

PS ZO

ZO

PM

PM

PS

ZO

ZO

ZO

PS

PS

PS

ZO

ZO

ZO

ZO

ZO

PM

PS

ZO

ZO

ZO

ZO

ZO

ZO

PB

ZO

ZO

ZO

ZO

ZO

ZO

ZO

4.3.3 Light Control Light energy is an important environmental factor for plant, which can directly affect plant’s growth and quality [19]. As shown in Fig. 4.1, the illumination is mainly controlled by top cover, side cover, and supplement light. As an artificial light source, light-emitting diodes (LEDs) can be used to promote vegetables growing faster in closed production systems, especially in the environments with insufficient light intensity [20]. In addition to sodium lamps, our research also used red and blue LED lamps, which provided variable spectral adjustment and energy-saving methods for plant growth and quality while planting [21]. In winter, the sunshine time is so short that the light needs to be supplemented during the day. The artificial light supplement scheme adopts the combination of LED and high-pressure sodium lamp. Compared with fluorescent lamps, it is highly energy efficient and it can increase the yield and quality of vegetables. The LEDs and the sodium lamps can be separately switched. Moreover, the daily light period and the dark period time can be set manually. During the bright period, depending on the effectiveness of photosynthesis, when the light is insufficient, the fill light can be switched on, and the turn-on time can be set by the operator. The greenhouse is located in the Antarctic, with polar day and polar night.

4 Research and Design on an Automatic Control System …

39

In the summer (polar day), the sunshine time is long, and it can be shaded by the top cover and the side cover to create a night effect, which is good for plant rest. On the contrary, in winter (polar night), the sodium lamps and the LED lamps are turned on to fill light and simulate the daylight to facilitate plant growth. The opening time and duration of the top cover and side cover are set according to the seasonal changes. This method of using artificial light sources to simulate day and night replacement and seasonal transformation is beneficial for plant growth.

4.3.4 Water and Fertilizer Control The greenhouse was divided into two areas, namely, the leafy vegetable area and the fruit vegetable area. Different cultivation racks were installed to plant leafy vegetables and fruit vegetables, respectively, and then a tidal irrigation scheme that is a kind of bottom water supply designed for nutrient solution cultivation and seedling cultivation was adopted. When irrigating, the water overflows from the water inlet and rises to a certain height in the culture frame, and then the crop absorbs water from the bottom by siphon principle. After the irrigation is completed, the water subsides from the water outlet, and the whole process is called “floor tide” and “ebb tide.” Tidal irrigation has the advantages of reducing disease, precise and easy irrigation, and saving water resources [22]. As shown in Fig. 4.6, because of the different irrigation schemes for leafy vegetables and fruit ones, two sets of irrigation devices were used. The nutrient tank was equipped with EC and PH meter to facilitate nutrient solution configuration and real-time monitoring. The timing and duration of the irrigation can be set as many times as needed with parameters, as shown in the example in Table 4.3. Fig. 4.6 Irrigation scheme for fruit and leafy vegetables

Pipeline

EC PH meter

Pump

Nutrition liquid bucket for leaf vegetable

Pipeline

EC PH meter

Pump

Nutrition liquid bucket for fruit vegetable

40

K. Lin et al.

Table 4.3 Irrigation plan

Irrigation time

Execution time

Irrigation time 1 (08:00)

Duration 1 (10 min)

Irrigation time 2 (09:00)

Duration 2 (15 min)

…

…

4.4 System Implementation and Application The environmental sensors produced by Xuzhou Deren Environmental Technology Co., Ltd. were selected for installation. All the sensors are shown in Table 4.4. The internal layout of the greenhouse environmental controller and the on-site installation effect of the sensor are shown in Fig. 4.7. In Fig. 4.7a, the environmental controller, Table 4.4 Executive information Device number

Name

Relay board number

Relay output channel number

1

Leaf vegetable pump

1

2

Circulating fan

1

1

3

Fruit vegetable pump

1

2

4

Air heating pump

2

0

5

Air heating four-way valve (open)

2

1

6

Air heating four-way valve (close)

2

2

7

Floor heating pump

2

3

8

Floor heating four-way valve (open)

2

4

9

Floor heating four-way valve (close)

2

5

10

Fruit vegetable sodium lamp

2

6

11

Fruit vegetable LED lamp

2

7

12

Leaf vegetable sodium lamp

2

8

13

Leaf vegetable LED lamp 2

14

Top cover (open)

2

10

15

Top cover (close)

2

11

16

Side cover (open)

2

12

17

Side cover (close)

2

13

18

CO2 control valve

2

14

0

9

4 Research and Design on an Automatic Control System …

(a) Greenhouse environment controller

(c) EC, PH meter

41

(b) Data collection box

(d) Sensors of temperature, humidity, photosynthetic and total radiation

Fig. 4.7 Controller and sensors in greenhouse

the switch signal input board, and the switch signal output board are installed in the weak electricity box. The sensors use the Modbus protocol to read a 4-bytes sample value from the register with the address and function code, and then calculate the actual physical quantity value by a formula. For example, the total radiation calculation formula is (30,000/Sensitivity) × (Sample value – 4)/16. There are 18 routes of actuators that need to be controlled in this system (as shown in Table 4.4). Each one relay board has 16 outputs, so two relay boards are required, and the remaining relay output channels can be used as a backup. The correspondence between the channel and actuator can be set by software parameters, and the initial values are shown in Table 4.4. If a channel fails, it can be switched to a remaining spare one and the correspondence is modified on the software to improve system redundancy and reliability. The input signal of the actuator is shown in Table 4.5. At the same time, the relationship between a device number and a channel number can also be set by one parameter. The software interface is shown in Fig. 4.8. Operators can intuitively learn the temperature of the inlet and outlet of the heating pipe, temperature and humidity of leaf vegetable and fruit vegetable area, photosynthetic efficiency, total radiation, ground temperature, CO2 , and other information. The software automatically controls the operation of each actuator according to the control strategy. To ensure reliability, a

42 Table 4.5 Input information of switch value in place

K. Lin et al. Device ID

Name

Switch input board number

Channel number

1

Top cover (open)

1

0

2

Top cover (close)

1

1

3

Side cover (open)

1

2

4

Side cover (close)

1

3

Fig. 4.8 Interface of control software

manual control mode is also provided in the system with the computer. The automatic control system software function modules are shown in Fig. 4.9. The system saves the running data every 10 min. For example, in 2018, each parameter has a total of about 49,000 data points. The monthly average values of parameters were counted. The statistics of the ambient temperature, pipe heating inlet and outlet temperatures are as shown in Fig. 4.10a. The average CO2 , photosynthetic efficiency, and total radiation changes are shown in Fig. 4.10b. The average fill time of fruit and leaf vegetables is shown in Fig. 4.10c, and the average temperature and humidity changes are shown in Fig. 4.10d. Since it is summer in the Great Wall Station from November to February every year, the ambient temperature is relatively high. Thus, the heating water temperature is relatively low, and the required light filling time is less. A long period of greenhouse production showed that planting result is good in the control of our system.

4 Research and Design on an Automatic Control System …

43

Antarctic greenhouse environment automatic control system software User Management

Environmental information management

Actuator management

Administrator

Measurement parameter setting

Board channel settings

General user

Measurement parameter query

Execution record query

Permission settings

Real-time monitoring of environmental information

Actuator status monitoring

Parameter management Image parameter Operating parameters Site information parameter Serial port parameter setting

Fig. 4.9 Functional modules for automatic control software

(a) Ambient temperature, pipe heating inlet and outlet temperature

(c) Supplementary lighting duration

(b) CO 2 photosynthetic effective and total radiation

(d) Average temperature and humidity

Fig. 4.10 Environmental information of greenhouse in Great Wall Station by month statistics in 2018

44

K. Lin et al.

4.5 Conclusion With less valuable experience being offered for natural light greenhouse control in the Antarctic, we have designed an environmental monitoring plan and control strategy suitable for the greenhouse of the Great Wall Station. It has realized comprehensive monitoring of greenhouse environment including temperature, humidity, photosynthetic efficiency, total radiation, CO2 , pipeline temperature, and plant images, etc. Meanwhile, it can effectively control the important factors of greenhouse such as light, temperature, water, and fertilizer. After installation of the system in the Great Wall Station, only one resident emergency doctor is responsible for vegetable cultivation in greenhouse. The long-term practical applications and experiments at the Great Wall Station showed that all subsystems work normally, and both leafy vegetables and fruit vegetables are planted well. The greenhouse automatic control system not only has been running stably but has become a strong guarantee for the vegetable supply of the Antarctic expedition team members, and then has solved the problem of eating fresh vegetables. This paper has provided technical methods and data accumulation for the greenhouse control in extreme environments. Acknowledgments This work was supported by Science and Technology Support Project of the Ministry of Science and Technology (Key Technology Research and Demonstration of Vegetable Cultivation under Extreme Antarctic Environment, 2014BAD05B05).

References 1. Mao, H., Jin, C., Chen, Y.: Research progress and prospect on control methods of greenhouse environment. Transa. Chin. Soc. Agricult. Mach. 49(2), 1–13 (2018) (in Chinese) 2. Qin, L., Lu, L., Shi, C., et al.: Implementation of IOT-based greenhouse intelligent monitoring system. Trans. Chin. Soc. Agricul. Mach. 46(3), 261–266 (2015) (in Chinese) 3. Fourati, F.: Multiple neural control of a greenhouse. Neurocomputing 139(1), 138–144 (2014) 4. Lijun, C.H.E.N., Shangfeng, D.U., Yaofeng, H.E., et al.: Robust model predictive control for greenhouse temperature based on particle swarm optimization. Inform. Process. Agricult. 5(3), 329–338 (2018) 5. Ma, D., Neal, C., Maki, H., et al.: Greenhouse environment modeling and simulation for microclimate control. Comput. Electr. Agricult. 162, 134–142 (2019) 6. Doaa, M., Hanaa, T.: Analysis and design of greenhouse temperature control using adaptive neuro-fuzzy inference system. J. Electr. Syst. Inform. Technol. 4(1), 34–48 (2017) 7. Wang, L., Zhang, H.: An adaptive fuzzy hierarchical control for maintaining solar greenhouse Temperature. Comput. Electr. Agricult. 155, 251–256 (2018) 8. Li, S., Li, M., Wang, X.: Design of greenhouse environment controller based on fuzzy adaptive algorithm. In: 27th Chinese Control and Decision Conference (CCDC), pp. 2644–2647 (2015) 9. Liang, M., Chen, L., He, Y., et al.: Greenhouse temperature predictive control for energy saving using switch actuators. IFAC Papers on Line 51-17, 747–751 (2018) 10. Revathi, S., Sivakumaran, N.: Fuzzy based temperature control of greenhouse. IFAC-Papers on Line 49(1), 549–554 (2016) 11. Tang, Y., Jia, M., Mei, Y., et al.: 3D intelligent supplement light illumination using hybrid sunlight and LED for greenhouse plants. Optik 183, 367–374 (2019)

4 Research and Design on an Automatic Control System …

45

12. Ma, F., Liu, H., Yang, S., et al.: Effects of ebb and flow irrigation on soilless culture potted anthurium. Trans. Chin. Soc. Agricult. Eng. 28(24), 115–120 (2012) (in Chinese) 13. Arias, A., Rodriguez, F., et al.: Multi-objective hierarchical control architecture for greenhouse crop growth. Automatica 48(3), 490–498 (2012) 14. Du, Y., Gu, X., Wang, J., et al.: Yield and gas exchange of greenhouse tomato at different nitrogen levels under aerated irrigation. Sci. Total Environ. 668, 1156–1164 (2019) 15. Li, J., Liu, H., Wang, H., et al.: Managing irrigation and fertilization for the sustainable cultivation of greenhouse vegetables. Agricult. Water Manag. 210, 354–363 (2018) 16. Sagrado, J., Sanchez, J.A., Rodriguez, F., et al.: Bayesian networks for greenhouse temperature control. J. Appl. Logic 17, 25–35 (2016) 17. Marco, A., Julio, C., et al.: Temperature control in a MISO greenhouse by inverting its fuzzy model. Comput. Electr. Agricult. 124, 168–174 (2016) 18. Xie, J., Gao, P., Mo, H., et al.: Design and optimization of intelligent irrigation decision system in litchi orchard based on fuzzy controller. Trans. Chin. Soc. Agricult. Mach. 49(8), 26–32 (2018) (in Chinese) 19. Singh, D., Basu, C.: LEDs for energy efficient greenhouse lighting. Renew. Sustain. Energy Rev. 49, 139–147 (2015) 20. Xu, Yi., Chang, Y., et al.: The research on LED supplementary lighting system for plants. Optik 127(18), 7193–7201 (2016) 21. Hytonen, T., et al.: Effects of LED light spectra on lettuce growth and nutritional composition. Light. Res. Technol. 50(6), 880–893 (2018) 22. Gu, S., Yang, Y., Zhang, Y., et al.: Development status of automated equipment systems for greenhouse vegetable seedlings production in Netherlands and its inspiration for China. Trans. Chin. Soc. Agricult. Eng. 29(14), 185–194 (2013) (in Chinese)

Chapter 5

Design of Microwave Triple-Frequency Multimeter Based on Multilayer MoS2 Zhilong Zhao, Xiaoling Zhong, Ting Ting Guo, Bing Wu, Li Wen, Liangyi Deng, and Yuting Jiang

Abstract In this paper, a microwave triple-frequency multiplier based on multilayer MoS2 is designed. The multilayer MoS2 stripped from the MoS2 crystal has a strong nonlinearity. It has the physical characteristics of developing a microwave multiplier, such as a tripler. A multilayer MoS2 film is used to cover the microstrip line gap, and a triple-frequency multiplier (MFT) fabricated by multilayer MoS2 is designed. If the input power of the signal source is 15 dBm and the bandwidth of the input frequency is 0.8–1.2 GHz, the minimum conversion loss through the MFT is − 41.5 dBm, and the even harmonic nonlinearity is far weaker than the odd harmonic nonlinearity. Therefore, multilayer MoS2 is a promising MFT material. Finally, by adding a recycling branch behind the tripler, the third-frequency output power is increased by recovering the low-order signal harmonics.

5.1 Introduction There are nonlinear signal sources and frequency multipliers in microwave circuits, and they are widely used in electronic systems such as radiometers, radars, and wireless communications. Multipliers used in the microwave and millimeter wave bands are usually designed with Schottky diodes [1–7]. In recent years, as people have discovered that two-dimensional materials such as Graphene and multilayer MoS2 have unique nonlinear characteristics, in microwave signal sources [8–10], mixers [11, 12], optical signal sources [13–15], and other aspects have broad application prospects, causing extensive research. In 2017, Antti Saynatjoki systematically studied the unique nonlinear properties of single and multilayer MoS2. Studies have shown that the single-layer MoS2 has good optical nonlinearity: the third harmonic intensity generated by the MFT designed by MoS2 is 30 times of the second harmonic intensity, and the wave intensity of the fourth harmonic intensity and the second harmonic is consistent. Z. Zhao (B) · X. Zhong · T. T. Guo · B. Wu · L. Wen · L. Deng · Y. Jiang Chengdu University of Technology, Chengdu 610000, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_5

47

48

Z. Zhao et al.

Studies have shown that single or multiple layers of MoS2 have strong optical nonlinearities [13–15]. However, little research has been done on the microwave nonlinear characteristics of single or multilayer MoS2. A microwave multilayer MoS2 frequency tripler (MFT) was developed and its output power and spectrum were tested. In order to study the microwave nonlinear characteristics of MoS2, an MFT was designed with multilayer MoS2, and its output power and spectrum were tested. In addition, the recycling section and recycling odd harmonics are increased to increase the output power of the MFT.

5.2 Model and Test Method 5.2.1 Molecular Structure Model of MoS2 The two-dimensional material multilayer MoS2 has good nonlinearity. Without external bias, the input voltage (u) and output current (i) show a nonlinear relationship, as shown in Eq. (5.1). At u = 0, without any external bias, the input voltage (u) and the output current (i) satisfies the nonlinear relationship as in Eq. (5.1) i = f (u) = a0 + a1 u + a2 u 2 + · · · + an u n + · · · =

∞ 

un

(5.1)

n=0

an (n = 0, 1, 2, . . .) is determined by Eq. (5.2).  1 d n f (u)  1 n = an = f (0)  n n! du n! u=0

(5.2)

The molecular structure of the multilayer MoS2 is shown in Fig. 5.1. Each molecular layer of MoS2 is composed of a hexagonal lattice of two S atoms and a Mo atomic sheet, and a hexagonal prism is positioned between the Mo atom sheet and the S atom sheet.

5.2.2 Test Methods Multilayer MoS2 has also been found to have good optical nonlinearity [13–15], and its microwave nonlinearity needs further confirmation. In order to verify its microwave nonlinear characteristics, an MFT based on multilayer MoS2 is designed. The flowchart is shown in Fig. 5.2. The design method of the MFT based on the multilayer MoS2 is composed of a microstrip line having a small gap in the middle, and the gap of the microstrip line is covered with a multilayer MoS2 film peeled off from the MoS2 crystal. The basic structure is shown in Fig. 5.3.

5 Design of Microwave Triple-Frequency Multimeter …

49

Fig. 5.1 Molecular structure of multilayer MoS2

Fig. 5.2 MFT block diagram

Fig. 5.3 Multilayer MoS2 covers microstrip line gap

The input and output ports of the MFT are connected to a microstrip line connected to both sides of the multilayer MoS2 film. The impedance of the microstrip line on

50

Z. Zhao et al.

Fig. 5.4 Microstrip structure of multilayer MoS2

a 0.127 mm thick Rogers 5880 substrate is 50 , and the substrate is at 1 GHz. The dielectric constant is ε = 2.2, and the thickness of the microstrip line is 0.035 mm. The specific structure is shown in Fig. 5.4.

5.3 Production and Testing 5.3.1 Experimental Method and Structural Analysis On the basis of above work, design and manufacture the MFT. The physical map of the MFT is shown in Fig. 5.5. The performance of the MFT was verified using the method shown in Fig. 5.6. First, the input fundamental signal is generated by the signal source, and then the white noise signal mixed by the fundamental signal

Fig. 5.5 Physical map of MFT

Fig. 5.6 MFT output spectrum test block diagram

5 Design of Microwave Triple-Frequency Multimeter …

51

is filtered through the filter to obtain a pure fundamental signal, thereby avoiding the harmonic interference of the signal generator to the tripler output. The impact of signal spectrum performance, then, the insertion loss of the coaxial cable (lines 1,2, and 3) is tested to accurately calculate the power output of the MFT. Then use the spectrum analyzer to measure the output power of the MFT, output triple frequency, and the output power of the double frequency, and plot it into a graph, as shown in Fig. 5.7. From the test results, it can be concluded that when the input power of the signal source is 15 dBm and the input frequency is 0.8–1.2 GHz, the maximum power of the corresponding MFT output is −27.5 dBm, and the corresponding minimum conversion loss is −41.5 dBm. The output double frequency power is completely less than −50 dBm. The MFT spectrum at 1 GHz input frequency is shown in Fig. 5.8. These performance results show that the output power of the even harmonic signal of the MFT is much smaller than the output power of the odd harmonic signal. From the above experimental results, it can be observed that the output power of the oddorder harmonic signal of the MFT is much larger than that of the even-order harmonic. Therefore, the nonlinear characteristic of the MFT designed by the multilayer MoS2 can be approximated as Eq. (5.3): I ≈ a1 u + a3 u 3 + · · · + a2n+1 u 2n+1 + · · · ≈

∞  n=0

Fig. 5.7 Test result graph

a2n+1 u 2n+1

(5.3)

52

Z. Zhao et al.

Fig. 5.8 MFT spectrum with input frequency of 1 GHz

5.3.2 Increase Recycling Section The MFT circuit designed by MoS2 outputs only the fundamental and odd harmonics, and the even harmonics are almost zero. The amplitudes of the fundamental and odd harmonics of the output are shown in Fig. 5.9. From the Fig. 5.9 we can conclude that the output power of the fundamental, third harmonic, and fifth harmonic signals is about 95% of the total power. Therefore, in order to enhance the output efficiency of the MoS2 triple frequency, it is necessary to increase the recovery section and recover the power of the lower harmonics to increase the output power of the MoS2 tripler. As shown in Fig. 5.10, the triple-frequency effect of the MoS2 is much better than that of the triple multiplier without increasing the recovery section. Therefore, increasing the recovery section can increase the output power of the MFT.

Fig. 5.9 Output harmonic signal power ratio

5 Design of Microwave Triple-Frequency Multimeter …

53

Fig. 5.10 Increase the frequency output power after recycling

5.4 Conclusions The MFT was tested by increasing the input power and frequency, and the test results were analyzed. The nonlinear characteristics of MoS2 were analyzed by the test results. Research shows that the multilayer MoS2 has good nonlinearity in the microwave band, and the output power of the even harmonic signal is much weaker than the output power of the odd harmonic. The minimum conversion loss of the multilayer MoS2 is −41.5 dB, and the MFT outputs only the fundamental and odd harmonics, and the even harmonics are almost zero. Therefore, the multilayer MoS2 may be a material for manufacturing the MFT in the future. Acknowledgment This work is supported by the National Natural Science Foundation of China, project number 41574137.

References 1. Archer, J.W.: Millimeter wavelength frequency multipliers. IEEE Trans. Microw. Theory Technol. 29, 552–557 (1998) 2. Raisanen, A.V.: Frequency multipliers for millimeter and submillimeter wavelengths. Proc. IEEE 80, 1842–1852 (1992) 3. Maiwald, F., et al.: Planar Schottky diode frequency multiplier for molecular spectroscopy up to 1.3 THz. IEEE Microw. Guided Wave Lett. 9, 198–200 (1999) 4. Guo, J., Xu, J., Qian, C.: A new scheme for the design of balanced frequency tripler with Schottky diodes. Progr. Electromagn. Res. 137, 407–424 (2013)

54

Z. Zhao et al.

5. Campos-Roca, Y., et al.: An optimized 25.5–76.5 GHz PHEMT-based coplanar frequency tripler. IEEE Microw. Guided Wave Lett. 10, 242–244 (2000) 6. Louhi, J.T., Raisanen, A.V.: Optimization of the Schottky varactor for frequency multiplier applications at submillimeter wavelengths. IEEE Microw. Guided Wave Lett. 6, 241–242 (1996) 7. Louhi, J.T., Raisanen, A.V.: On the modeling and optimization of Schottky varactor frequency multipliers at submillimetre wavelengths. IEEE Trans. Microw. Theory Tech. 43, 922–926 (1995) 8. Mikhailov, S.A.: Non-linear electromagnetic response of grapheme. Europhys. Lett. 79 (2007) 9. Camblor, R., et al.: Microwave frequency tripler based on a microstrip gap with graphene. J. Electromagn. Waves Appl. 25, 1921–1929 (2011) 10. Fang, Y., et al.: Graphene frequency tripler design using reflector networks. IEICE Electron. Exp. 15, 1–7 (2018) 11. Hotopan, G.R., et al.: Millimeter wave microstrip mixer based on Graphene. Progr. Electromagn. Res. 118, 57–69 (2011) 12. Hotopan, G.R., et al.: Millimeter wave subharmonic mixer implementation using Graphene film coating. Progr. Electromagn. Res. 140, 781–794 (2013) 13. Li, Y., et al.: Probing symmetry properties of few-layer MoS2 and h-BN by optical secondharmonic generation. Nano Lett. 13, 3329–3333 (2013) 14. Janisch, C., et al.: Extraordinary second harmonic generation in tungsten disulfide monolayers. Sci. Rep. 4, 5530 (2014) 15. Säynätjoki, A., et al.: Ultra-strong nonlinear optical processes and trigonal warping in MoS2 layers. Nat. Commun. 8, 893 (2016)

Chapter 6

Correlation Research on the Structure of the Apple Tree Vigor and Its Fruit Quality Zhijun Wang, Peng Lan, and Fenggang Sun

Abstract Various factors will affect the fruit quality of apple, mainly environmental factors, soil nutrient, environmental humidity, and tree’s structure. This paper researches on the correlation between the affecting indicators (including crown diameter, crown height, trunk girth, and tree crown volume) and the fruit quality (including fruit hardness, soluble sugar content, and soluble solids). From these correlations, we can conclude that: (1) The correlation between fruit hardness and the crown diameter, the trunk girth and the crown height is negative; (2) the correlation between the soluble sugar and natural logarithm of the trunk girth is negative; and (3) the correlation between the soluble solids and the trunk height is positive, and the correlation between soluble solids and crown diameter and volume is negative.

6.1 Introduction Apple is cultivated as one of the main fruit trees in the world. In the year 2017, the total area harvested in the world for apples is 4.9 million hectares, and the total production of apples is 83.1 million tons. Meanwhile, the total area harvested in China is 2.22 million hectares and the total production is 41.39 million tons, accounting for 45.3 and 49.8% of the world’s total, respectively [1]. As one of the most important apple producing areas in China, Shandong province produces 9.395 million tons of apples in 2017, accounting for 22.70% of the total apple production. Shandong province is located in the north latitude between 32° and 40°, where the climate belongs to the temperate zone with humid air, abundant light, and large the temperature difference between day and night. These factors help to form the effective accumulation of sugar apple. Moreover, there are a large number of hills Z. Wang (B) · P. Lan · F. Sun College of Information Science and Engineering, Shandong Agricultural University, Tai’an 271018, China e-mail: [email protected] F. Sun e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_6

55

56

Z. Wang et al.

in Shandong province, where the soil is rich in phosphorus, potassium, calcium, magnesium, and other elements. The superior geographical location and natural conditions make Shandong province very suitable for apple planting. The quality of apples is one of the key factors to determine the competitiveness of apples in markets. Therefore, it is of great importance to improve the apple quality [2]. There exist many different factors to measure apple quality, and these factors are often related and relatively independent [3]. In [3], Cheng et al. applied the principal component analysis and cluster analysis to determine the main factors that affect the quality. The authors in [4] use the meteorological and phonological monitoring data to analyze the apple quality. In [5], the correlation of apple quality from different habitants with the geographical coordinates are analyzed, proving that the growing location of the same apple breed affect the fruit quality. Lu et al. [6] study the relationship between growth status of apple and soil condition deeply to guide fertilization scientifically. Then in [7], Jiang et al. proposed to evaluate the apple quality by analytic hierarchy process combined with sensory evaluation. Also, Ma et al. proposed to jointly use the analytic hierarchy process and gray relational grade analysis to evaluate the fruit quality [8]. In this paper, we use the regression approach to research on the correlation between indicators of crown diameter, crown height, trunk girth, tree crown volume, leaves quantity, and fruit quality in terms of fruit hardness, soluble sugar content, and soluble solids. We then predict the fruit quality based on the regression results to verify the effectiveness.

6.2 Materials and Experiments The experiment was carried out in 49 orchards of 21 regions in Shandong province. The apple variety is Red Delicious, and the age is about 13 years. The soil is acid sandy soil and the pH maintains at 5.24–7.03. The fruit size and color of apple trees selected in each orchard are more uniform and the yield is more stable.

6.3 Results and Analysis 6.3.1 Correlation Between Indicators of Crown Diameter, Crown Height, Trunk Girth, and Hardness of the Fruit 6.3.1.1

Correlation Between Crown Height, Trunk Girth, and Hardness

Take the fruit hardness (y) as dependent variable, crown height (x1 ) and trunk girth (x2 ) as independent variables. The regression equations can be achieved based on

6 Correlation Research on the Structure of the Apple Tree Vigor … Table 6.1 Regression statistics of crown height, trunk girth, and fruit firmness

57

Regression statistics Multiple R

0.585996

R square

0.343392

Adjusted R square

0.314843

Standard error

0.591865

Observed value

49

the data from 49 orchards. It has been proved that the regression equation and its parameters passed the test. The results are shown in Tables 6.1, 6.2 and 6.3. From Table 6.3, the regression equation about crown height, trunk girth, and fruit hardness can be given as follows: y = 12.22284 − 0.00679x1 − 0.02256x2

(6.1)

Use this regression equation to predict the fruit hardness from the 49 samples, and the results are shown in Table 6.4. The average error between the predicted values and the measured values of the fruit hardness from Table 6.4 is 4.74%. Therefore, the prediction accuracy is high.

6.3.1.2

Correlation Between Trunk Girth, Crown Diameter, and Hardness

Take the fruit hardness (y) as the dependent variable, trunk girth (x1 ) and crown diameter (x2 ) as the independent variable, the regression equations can be achieved based on the data from 49 orchards. The results are shown in Table 6.5. From Table 6.5, the regression equation about the trunk girth, the crown diameter, and the fruit hardness is: Table 6.2 Variance analysis of crown height, trunk girth, and fruit firmness Source Regression analysis

Degree of freedom 2

Squares sum 8.427252

Residual error

46

16.11397

Total

48

24.54122

Mean square

F value

Sig

4.213626

12.02849

6.28E-05

0.350304

Table 6.3 Parameter estimation of fruit firmness (crown height, trunk girth) Coefficients

Standard error

t Stat

P-value

Intercept

12.22284

0.681228

17.94236

2.12E-22

Crown height (cm)

−0.00679

0.00208

−3.26215

0.002087

Trunk girth (cm)

−0.02256

0.00748

−3.01644

0.004157

58

Z. Wang et al.

Table 6.4 Accuracy tests of the fruit firmness (crown height, trunk girth) No.

Fruit firmness Predicted value (kg/cm2 )

Error (%)

No.

Measured value (kg/cm2 )

Fruit firmness Predicted value (kg/cm2 )

Error (%)

Measured value (kg/cm2 )

1

8.60

8.30

3.54

26

9.03

9.10

0.74

2

8.50

7.90

7.04

27

9.03

10.20

9.41

3

8.33

8.20

1.60

28

9.32

8.60

9.34

4

8.17

8.10

0.81

29

9.49

9.90

2.67

5

8.18

9.00

10.05

30

9.64

9.60

0.53

6

8.77

8.80

0.37

31

9.65

9.20

3.02

7

8.74

8.60

1.61

32

9.49

9.50

0.91

8

8.50

8.40

1.16

33

9.59

9.40

1.63

9

8.43

8.40

0.39

34

9.25

9.60

2.54

10

8.69

9.10

4.71

35

9.36

9.50

6.58

11

8.63

8.00

7.27

36

8.91

9.80

8.22

12

9.28

8.70

6.21

37

9.06

9.20

1.89

13

8.73

9.80

12.21

38

9.03

8.60

6.51

14

8.51

8.50

0.09

39

9.20

8.10

10.21

15

9.00

8.80

2.18

40

9.02

8.10

10.82

16

9.30

9.00

3.24

41

9.08

8.90

0.82

17

9.74

9.40

3.48

42

8.97

8.60

1.34

18

9.28

8.90

4.07

43

8.72

8.30

2.53

19

8.92

9.20

3.18

44

8.52

8.30

2.78

20

9.40

9.50

1.01

45

8.54

8.00

2.99

21

8.91

10.20

14.48

46

8.25

8.00

10.55

22

9.29

10.90

17.28

47

8.94

8.30

2.74

23

9.34

10.40

11.36

48

8.53

8.30

0.84

24

9.26

9.70

4.73

49

8.37

8.10

10.14

25

9.33

9.30

0.27

Table 6.5 Parameter estimation of the fruit firmness (trunk girth, crown diameter) Intercept

Coefficients

Standard error

t Stat

P-value

11.53938

0.638809

18.0639

1.62E-22

Trunk girth (cm)

−0.00638

0.011814

−0.53977

0.59196

Crown diameter (cm)

−0.00482

0.002093

−2.3033

0.025833

6 Correlation Research on the Structure of the Apple Tree Vigor …

59

y = 11.53938 − 0.00638x1 − 0.00482x2

(6.2)

Use this regression equation to predict fruit hardness from 49 samples, and the results are shown in Table 6.6. In Table 6.6, the average error between the predicted values and the measured values of fruit hardness is 5.57%. Table 6.6 Accuracy test of the fruit firmness (trunk girth, crown diameter) No.

Fruit firmness Predictive value (kg/cm2 )

Error (%)

No.

Measured value (kg/cm2 )

Fruit firmness Predictive value (kg/cm2 )

Error (%) Measured value (kg/cm2 )

1

8.60

8.30

5.51

26

9.03

9.10

3.39

2

8.50

7.90

12.22

27

9.03

10.20

7.96

3

8.33

8.20

4.51

28

9.32

8.60

4.40

4

8.17

8.10

4.69

29

9.49

9.90

7.56

5

8.18

9.00

4.62

30

9.64

9.60

1.59

6

8.77

8.80

4.12

31

9.65

9.20

2.52

7

8.74

8.60

6.26

32

9.49

9.50

2.75

8

8.50

8.40

8.75

33

9.59

9.40

3.10

9

8.43

8.40

0.73

34

9.25

9.60

4.43

10

8.69

9.10

6.91

35

9.36

9.50

9.52

11

8.63

8.00

10.21

36

8.91

9.80

12.00

12

9.28

8.70

5.47

37

9.06

9.20

10.50

13

8.73

9.80

8.18

38

9.03

8.60

1.97

14

8.51

8.50

4.31

39

9.20

8.10

9.02

15

9.00

8.80

6.18

40

9.02

8.10

11.35

16

9.30

9.00

4.03

41

9.08

8.90

0.80

17

9.74

9.40

1.38

42

8.97

8.60

1.06

18

9.28

8.90

4.53

43

8.72

8.30

2.17

19

8.92

9.20

0.73

44

8.52

8.30

1.91

20

9.40

9.50

0.87

45

8.54

8.00

3.15

21

8.91

10.20

10.46

46

8.25

8.00

9.22

22

9.29

10.90

13.27

47

8.94

8.30

5.54

23

9.34

10.40

11.01

48

8.53

8.30

5.53

24

9.26

9.70

4.01

49

8.37

8.10

6.82

25

9.33

9.30

1.8

60

Z. Wang et al.

Table 6.7 Parameter estimation of soluble sugar (trunk girth) Coefficients

Standard error

Intercept

22.14433

2.982293

trunk girth(cm)

−2.01809

0.739463

t-value 7.425271 −2.72913

P-value 1.86E-09 0.008906

6.3.2 Correlation Research Between the Trunk Girth and the Soluble Sugar With the soluble sugar (y) as dependent variable and natural logarithm of trunk girth (ln x) as independent variable, the regression equation is achieved based on the data from 49 orchards. The results are shown in Table 6.7. The regression equation about soluble sugar and trunk girth is: y = 22.14433 − 2.01809 ln x

(6.3)

Use this regression equation to predict the soluble sugar from 49 samples, and the results are shown in Table 6.8. The average error between the predicted values and the measured values of soluble sugar is 6.19%.

6.3.3 Correlation Research Between the Structure of Tree Vigor and Soluble Solids This section studies the correlation between trunk height (cm), the crown diameter (cm), the tree crown volume (m3 /mu), and the soluble solids (%). The research objects are the samples with the soluble solids content being higher than 13.5%.

6.3.3.1

Correlation Between the Trunk Height and the Soluble Solids in Fruit

Take the soluble solids (y) as dependent variable, trunk height (x) as independent variable. The regression equation can be achieved based on the samples with the soluble solids content being higher than 13.5%. The results are shown in Table 6.9. According to the Table 6.9, the regression equation is as follows: y = 13.0362 + 0.1942x

(6.4)

Use the Eq. (6.4) to predict the samples, and the results are shown in Table 6.10. The average error between the predicted values and the measured values of soluble solids from Table 6.10 is 4.17%.

6 Correlation Research on the Structure of the Apple Tree Vigor …

61

Table 6.8 Accuracy test of the soluble sugar (trunk girth) No.

Soluble sugar

Error (%)

Predictive value

Measured value

1

13.82

13.84

0.20

2

13.75

13.08

5.16

3

13.63

12.26

4

13.57

12.89

5

13.54

12.34

6

13.72

12.83

7

13.82

12.11

8

13.66

13.22

3.32

9

13.78

14.73

10

13.69

12.76

11

13.95

12 13

No.

Soluble sugar

Error (%)

Predictive value

Measured value

26

14.55

15.34

5.11

27

14.51

15.55

6.73

11.18

28

14.25

15.65

8.93

5.31

29

14.37

13.23

8.67

9.71

30

14.42

14.35

0.48

6.92

31

14.37

14.46

0.60

14.08

32

14.65

16.29

10.08

33

14.02

14.94

6.15

6.40

34

14.09

14.76

4.49

7.30

35

13.49

14.96

9.83

13.24

5.35

36

13.49

14.60

7.61

14.51

15.30

5.17

37

13.25

13.42

1.24

13.92

13.73

1.36

38

14.33

11.97

19.75

14

13.78

12.63

9.15

39

13.95

14.65

4.78

15

14.09

14.00

0.70

40

13.75

14.03

1.96

16

14.21

15.34

7.37

41

13.85

14.06

1.52

17

14.55

13.92

4.56

42

13.40

11.54

16.12

18

14.42

13.58

6.13

43

13.35

15.41

13.36

19

14.37

14.88

3.37

44

13.63

14.24

4.29

20

14.51

13.21

9.82

45

13.30

14.98

11.19

21

14.25

13.87

2.72

46

14.25

14.11

0.95

22

14.46

13.83

4.58

47

14.09

15.05

6.35

23

14.55

15.23

4.43

48

13.57

12.98

4.59

24

14.37

14.71

2.26

49

14.17

14.94

5.13

25

14.70

13.75

6.92

Table 6.9 Parameter estimation of soluble solids (trunk height) Coefficients Intercept Trunk height (cm)

13.0362 0.01942

Standard error

t Stat

0.344562

37.83411

0.007606

2.553248

P-value 2.67E-16 0.022055

62

Z. Wang et al.

Table 6.10 Accuracy test of the soluble solids (trunk height) No.

Predictive value

Measured value

Residuals

14

13.2984

13.50

0.2016

1.49

40

13.2984

13.50

0.2016

1.49

42

13.2984

13.50

0.2016

1.49

10

13.3003

13.60

0.2997

2.20

34

13.3023

13.70

0.3977

2.90

35

13.3023

13.70

0.3977

2.90

7

13.3042

13.80

0.4958

3.59

8

13.3042

13.80

0.4958

3.59

9

13.3042

13.80

0.4958

3.59

41

13.3061

13.90

0.5939

4.27

5

13.3081

14.00

0.6919

4.94

6

13.3081

14.00

0.6919

4.94

13

13.3081

14.00

0.6919

4.94

33

13.3081

14.00

0.6919

4.94

24

13.3100

14.10

0.7900

5.60

17

13.3139

14.30

0.9861

6.90

30

13.3275

15.00

1.6725

11.15

6.3.3.2

Error (%)

Correlation Between the Trunk Height and the Soluble Solids in Fruit

Following the similar approach, we can obtain the correlation between crown diameter (x) and soluble solids (y) as y = 15.4659 − 0.00347x

(6.5)

The correlation between crown volume (x) and the soluble solids (y) is y = 14.40141 − 0.0033x

(6.6)

6.4 Conclusions The correlation analysis between crown diameters, height, trunk girth, and fruit hardness shows that fruit hardness is relevant to the crown diameter, the trunk girth, and the crown height, and from the three corresponding regression equations, there is a negative correlation between the fruit hardness and the three factors. Therefore,

6 Correlation Research on the Structure of the Apple Tree Vigor …

63

reasonable control of the crown height, diameter, and the trunk girth has a positive significance to improve the fruit hardness. When forming the regression equations, the three tree structure indicators (the crown diameter, height, and the trunk girth), each has a significant correlation with the fruit hardness. But due to non-normality and interactive dependence of the sampling data, regression equation between the fruit hardness and a combination of these three factors cannot be easily obtained. The relation between the apple soluble sugar and the trunk girth shows that the two could meet the following relationship: y = 22.14433 − 2.01809 ln x. When soluble sugar content equals 14%, the trunk girth is calculated as 56.5784 cm from the regression equation. Owing to the negative correlation between the soluble sugar and natural logarithm of the trunk girth, if we choose an orchard with the soluble sugar content more than 14% as a high-quality one, the average trunk girth should be between 56.5784 ± 2σ (σ = 1.099193 cm). The correlation between trunk height, crown diameter, tree crown volume, and the soluble solids shows that there exists a positive correlation between the soluble solids and the trunk height. The increase of trunk height may keep trees ventilated and transparent to light which is conducive for the formulation of the soluble solids. The soluble solids in fruits have a negative correlation with the crown diameter and the tree crown volume. Proper reduction of the tree crown volume and the crown diameter is conducive for the formulation of the soluble solids.

References 1. Apple production in 2017; Crops/World Regions/Production Quantity. FAOSTAT, UN Food & Agriculture Organization, Statistics Division. 2017. Archived from the original on 11 May 2017 2. Du, J., Li, X., Shi, H.: Current situation and prospect of our apple industry. J. Hebei Agric. Sci. 10, 97–101 (2006). (in Chinese) 3. Wei, Q., Cheng, S., Ding D.: Factors selection to evaluate apple quality. China Fruits 4 (1997). (in Chinese) 4. Qu, Z., Zheng, X., Liu, L., Zhang, Y.: Relationship between apple quality and meteorological factors in different ecological regions in Shaanxi. Meteorol Monthly 43, 872–878 (2017). (in Chinese) 5. Li, Z., Guo, Y., Sun, L., et al.: Quality differences of “Nagafu 2” apple from different habitats and its correlation with geographical coordinates. J. Shaanxi Normal Univ. (Natural Science Edition) 40(2012), 98–103 (2012). (in Chinese) 6. Lu, C., Xue, X., Wang, C., et al.: Correlation analysis on fruit quality and leaves nutrition and soil nutrient in apple orchard of shandong province. Chinese Agric. Sci. Bull. 27, 168–172 (2011). (in Chinese) 7. Jiang, X., Ma, J., Xu, J., et al.: Establishment and verification of an evaluation model in apple quality. J. Gansu Agric Univ 54, 69–77 (2019). (in Chinese) 8. Ma, Q., Liang, L., Li, Q., et al.: Synthetical evaluation of the fruit quality of ‘Dongzao’ advanced selections using analytic hierarchy process and grey relational grade analysis. Acta Horticu 940, 213–220 (2012)

Chapter 7

Real-Time Data Transmission Design of Unmanned Aerial Vehicle Gamma Spectrometer Xiaoling Zhong, ZhiLong Zhao, Yuting Jiang, Bing Wu, MingHao Guo, and Yong Fang Abstract Unmanned aerial vehicle gamma spectroscopy is an important method for conducting nuclear emergency and rapid geological prospecting, and the design of real-time data transmission is a key technical difficulty. In this paper, four 2L large volume NaI(TL) gamma spectrometers are designed to improve the sensitivity of ground detection. GPS second pulses are used for time synchronization and added as time stamp markers to the measured energy spectrum data. The spectral line data is transmitted to the airborne industrial computer under the TCP/IP transmission protocol, and the spectral line data is packed together with other information, and then the LZMA compression algorithm is used to reduce the volume of the packed data. The packaged data and the high-definition HDMI image are sent to the ground through the digital transmission and transmission machine. The self-written RTCP transmission protocol is used in the transmission process to ensure the reliability and real-time performance of the data. Finally, the line display and the radionuclide analysis are realized on the ground receiving end, thereby ensuring the time precision and the real-time performance of long-distance transmission. The real-time data transmission method designed and developed in this paper can be widely applied to the UAV radiation measurement occasion, which can solve the problem of real-time transmission of radiation measurement data.

7.1 Introduction The measurement of gamma-ray spectrum on UAV is to measure the specific energy of the crustal rocks or radioactive sources by gamma-ray spectrometer during the flight of UAV. According to the gamma-ray spectrum transmitted to the ground by UAV, the distribution characteristics of gamma-ray irradiation rate or radioactive elements in the area are obtained [1]. At present, this technology is mainly used in the field of environmental radiation monitoring, especially in the field of environmental X. Zhong (B) · Z. Zhao · Y. Jiang · B. Wu · M. Guo · Y. Fang Chengdu University of Technology, Chengdu 610059, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_7

65

66

X. Zhong et al.

radioactivity evaluation, aviation nuclear emergency, and the search for radioactive minerals. With the increasing use of unmanned aerial vehicle gamma spectroscopy, there are also problems such as low signal strength due to long-distance transmission and severe network packet loss. These problems make it impossible to transmit data to the ground in real time during flight. Therefore, this paper designs a real-time transmission system for unmanned aerial vehicle gamma spectroscopy measurement data that can transmit spectral line data and image information in real time within 40 km, and can display γ energy spectrum and high-altitude image on the ground display interface. The system uses four large volume NaI(TL) crystals to increase the instantaneous counting rate to improve detection efficiency, and uses GPS second pulse as time reference to improve data synchronization. Lempel-Ziv– Markov chain algorithm (LZMA) is used in data transmission. The compression algorithm greatly compresses the spectral line data to improve the transmission efficiency. At the same time, it uses TCP/IP transmission protocol and self-written UDP-based reliable transmission control protocol (RTCP) to solve packet loss retransmission and disconnected storage in data transmission. HDMI format is selected for efficient transmission of HD image information. Through the above measures, the spectral line data and the image are simultaneously transmitted, and the γ energy spectrum is displayed in real time on the ground display interface.

7.2 Data Transmission Overall Structure Design Figure 7.1 shows the data transmission flow diagram of the unmanned aerial vehicle gamma spectroscopy measurement system. As shown in the figure, the nuclear pulse signal amplified by the photomultiplier tube is converted into a digital signal by the digital multi-channel DMCA-1003 independently developed by Chengdu University of Technology, and the spectral line data is sent to the industrial computer through the POE router. The industrial computer receives the radar height, the Geiger count, the attitude information, and the GPS information from the serial port, and the industrial computer program packs and compresses them. The packaged data is sent to the digital transmission host through the network port, and the host sends the packaged data to the ground digital transmission and transmission slave by way of radio broadcast. Then the slave sends the received data to the ground control platform for decompression. The end achieves stable spectrum through the 40 K characteristic peak [2] and realizes nuclide identification [3], displaying functions such as GIS map. In order to obtain a large instantaneous count rate in a short time, the system uses four 2L large volume NaI(TL) scintillator detectors to increase the instantaneous count rate. The digital multi-channel DMCA-1003 rapidly converts the nuclear pulse signal into a digital signal. The DMCA-1003 is a self-developed device developed by Chengdu University of Technology. It has a full set of development technology and intellectual property, as well as technical and conceptual leading line pre-processing command settings. This setting can send the line prefetch command first when the industrial

7 Real-Time Data Transmission Design of Unmanned Aerial …

67

Fig. 7.1 Airborne gamma spectrum measurement system data transmission flow diagram

computer wants to acquire the multi-channel spectrum data, and quickly switch the ping-pong buffer internally after receiving the command in multiple channels. When the industrial computer sends the acquisition line command to obtain multiple data, the pre-processing command of the channel eliminates multiple dead time problems due to data transmission uncertainty, and is ideal for fast-moving measurement applications. In the working mode, the number of multi-channel energy channels is set to 2048 channels, and the system measurement period is set to 1 time/s. Then, a single multi-channel has 8 KB data in each cycle. When four multi-channels work simultaneously, each data transmitted during the cycle has 32 KB. Owing to the long-distance transmission and complex geographical environment, real-time data transmission and display of γ energy spectrum maps require high reliability and real-time performance.

7.2.1 Time Base The time base is used to trigger the simultaneous acquisition of four digitized multichannels and other information to ensure the time synchronization of the data information. The system uses the BX220 positioning board produced by Zhonghaida Company to output GPS second pulse (PPS [4]) as the time reference. The board has 20 ns timing accuracy and centimeter-level positioning output, which is a low-power

68

X. Zhong et al.

and high-precision navigation. The board has the advantages of fast positioning speed and high positioning accuracy. In the automatic measurement phase of the system, the frequency of receiving GPS information is set once/s as a time reference to issue commands to acquire data information of four digitized multi-channels and other devices.

7.2.2 Digital Multi-channel Data Transmission After the four digitized multi-channels acquire the spectral line data, the spectral line data is transmitted to the industrial computer for processing, and in this process, the data synchronous transmission and transmission reliability are guaranteed. In the process of transmitting the data of four digital multi-channel pulse amplitude analyzer to the industrial computer, a router needs to be added. The main reasons are as follows: First, there is only one network port of the industrial computer, which cannot meet four digital multi-channels and is connected to the industrial control at the same time. Second, because the digital multi-channel DMCA-1003 uses RJ45 network cable for data transmission and uses power over ethernet (PoE) to supply power, it needs to solve the problem of power supply. Therefore, the system uses the EdgeRouter PoE five-port router-type manufactured by UBIQUITI as the data relay station. The EdgeRouter PoE five-port router has four data input ports and one data output port. The EdgeRouter PoE five-port router can also provide four data ports. The data input interface is powered by 48 V to meet the requirements of PoE power supply. By adding this router, the function of acquiring data commands and transmitting spectral lines can be simultaneously transmitted to realize four multichannel pulse amplitude analyzers, thereby improving the synchronization of data acquisition time and reducing time error. In order to ensure that four multi-line data can be transmitted to the industrial computer with high efficiency and reliability, it needs to be secured by appropriate transmission path and communication protocol. The system is guaranteed by network cable and TCP/IP transmission communication protocol. The transmission rate of the network cable is 100 Mbps, and the transmission rate per second can reach 1000 KB or more, which satisfies the requirement of spectral data transmission speed of 32 KB/s. TCP/IP is the most widely used and reliable communication protocol, and TCP/IP internal timeout retransmission measures can ensure real-time data transmission. Each data passing through the digital multi-channel pulse amplitude analyzer and the router to the industrial computer is transmitted by the network cable. The data transmission of the digital multi-channel pulse amplitude analyzer and the industrial computer adopts the TCP/IP [5] transmission control protocol to ensure the reliability of the transmission. TCP/IP has an efficient error detection mechanism to ensure the reliability of data transmission. The IPC establishes the TCP server to set the IP address, and the router uses the DHCP method to assign each digital multi-channel IP address to ensure that the IPC and the multi-channel pulse amplitude analyzer

7 Real-Time Data Transmission Design of Unmanned Aerial …

69

work in the same IP segment, so that the IPC and the multi-channel establish a reliable connection. The TCP/IP protocol achieves data transmission reliability through serial number, checksum, acknowledgement response signal, retransmission control, connection management, window control, flow control, and congestion control. The error detection process is as follows: (1) Improve reliability by serial number and acknowledgment response signal: In TCP, when the receiving host receives the data packet from the client host, the receiving host returns an acknowledgment response signal of the received message. TCP achieves reliable data transmission by affirmative acknowledgement signal ACK packets. When the sender sends the data, it will wait for the confirmation reply from the peer. If there is a confirmation response, the data has successfully arrived at the peer host, and the sender deletes the data in the buffer. Conversely, if the acknowledgment response is not received, the possibility of packet loss is relatively large and retransmission is required. (2) Resend the data to ensure complete data transmission: During the set waiting time interval, if the sending host still does not receive an acknowledgment from the receiving end, it is considered that the data has been lost and needs to be resend. Therefore, according to the serial number and the acknowledgment response signal, the TCP/IP protocol can reliably transmit the spectral line data to the industrial computer.

7.2.3 IPC Program Data Processing After the spectral line data is transmitted to the industrial computer, it needs to be packaged and sent to the ground receiving end together with the GPS information and temperature information. Therefore, the gamma spectrum measurement program, drone gamma spectrum measurement program (DGSMP), based on the Visual Studio 2012 development environment is written in the industrial computer. DGSMP is written in C++ language and works in multi-thread [6] mode. The main functions are: receiving and forwarding data acquisition commands; receiving spectral line data and GPS information, and packaging and compressing them, and then sending them to the digital transmission host. Create an SQLite database to save data and more. After the data is packaged, the data needs to be compressed for the following reasons: (1) The volume of the transmitted data is reduced; the number of unpacking is reduced; the transmission time is shortened; the receiving time is shortened; and the probability of packet loss is reduced, so as to improve the reliability of longdistance transmission. (2) Reduce the proportion of the transmission bandwidth of the transmitted data in the digital transmission and transmission of the host, so that the image and the data can be transmitted to the ground without affecting each other. The system uses the LZMA [7] compression algorithm to compress the data. LZMA has high compression ratio, decompression takes up less memory, and has faster decompression. It is very suitable for embedded systems. In the automatic measurement mode, a fixed size of 32.22 KB data is sent to the ground. The compression

70

X. Zhong et al.

Table 7.1 Measuring system’s ability to compress data at different transmission times Transmission time/h

Volume before data compression/KB

Volume after data compression/KB

Compression ratio/%

1/60

32.23

5.44

16.90

1/6

32.23

7.01

21.73

0.5

32.23

8.27

25.65

1

32.23

9.07

28.05

2

32.23

9.40

29.19

5

32.23

11.01

34.16

effect test is shown in Table 7.1. The measurement system tests the data compression capability at different transmission times. It can be seen from the test results that as time increases, the volume after compression gradually increases, the compression ratio gradually increases, and the compression effect gradually decreases. This is because the amount of data in the previous period is small, the count is small, and the high bit of data is large. As the measurement time increases, the line count increases, the data high bit is filled, the volume is increased after being compressed by LZMA, and finally stabilizes at 11.01 KB. The compression ratio can reach up to 34.16%.

7.2.4 Data Transmission Between Drones and Ground Control Platforms The data transmission between the drone and the ground control platform consists of two parts: the packed compressed data (including spectral line data, radar height, Geiger count, attitude information, and GPS) and the image information of the HD camera are packaged in the industrial computer on the drone. In the data transmission of the industrial computer and the ground control platform, there are mainly longdistance transmission signals with fast attenuation, easy data loss and disconnection, and so on. For this reason, the data transmission equipment of this system adopts the DDL mini-type produced by Shenzhen Huaxiasheng Company. The transmission machine mainly solves the problem of weak signal strength and simultaneous transmission of data images. Its main advantages are as follows: (1) signal enhancement is large; signal receiving sensitivity is as high as −95 dBm; when transmitting data in the form of radio broadcast, it can achieve data within 50 km; (2) uses two channels for data and image transmission with a transmission speed of up to 30 Mbps; (3) power as low as 1 W. The image data collected by the HD camera is transmitted to the digital transmission and transmission machine in high-definition multimedia interface (HDMI) [8] format, and the data format is 1080 * 720 p, 30 frames/s. In order to ensure that the packed data can be sent to the ground in real time under abnormal network conditions, the system uses the self-written UDP-based

7 Real-Time Data Transmission Design of Unmanned Aerial …

71

reliable transmission control protocol (RTCP) to solve the problems of packet loss retransmission and disconnected storage. RTCP is a transport protocol based on the Connext DDS [9] solution. Connext DDS is a standard-based, open architecture that connects devices from deep embedded real-time platforms to servers across a wide range of networks while providing large data sets with microsecond performance and granular quality service control. Reliability and other quality of service (QOS) guarantees, especially for large data volumes, based on various transmissions (including packet-based transmission, unreliable network, multiplex, burst or high-latency transmission). High-speed transmission has significant advantages in real-time communication. Working principle: As shown in the left side of Fig. 7.2, a packaged data is split into three small packets and sent to the send buffer send queue. The receiver must receive three small packets of data to be a complete packaged data before it is uploaded to the application. Floor: Each line of the sending queue consists of three parts. The number 1 on the left indicates the queue number. The increment of the sequence number is automatically incremented by the system. The middle A is the data content, and the right X is the success identifier.

Fig. 7.2 RTCP flowchart

72

X. Zhong et al.

DATA: This contains the serial number and data content of the transmitted data. HB: This informs the ground receiving end of the data sequence number that should be received, and requires the ground receiving end to return the acknowledged received ACKMSG information. ACKMSG: The message returned by the data receiving end to the sending end. If the receiving queue has received all the data, it is accompanied by 0. If the receiving queue lacks certain serial number data, it will return the missing serial number to notify the sending end to resend. The communication flow during packet loss retransmission is shown in Fig. 7.2. When transmitting, DATA(1, A) carries queue number 1 and data content A, and will also be accompanied by a HB(1) message, and HB(1) will notify the ground receiving. The data that should be received at the same time requires the ground receiving end to return the acknowledged received ACKMSG information. If the DATA(1, A) network is lost, when the sending DATA(2, B) is correctly received and placed in the receiving queue, the receiving success identifier of the receiving √ queue number 2 is changed from X to , and the receiving end passes HB (1–2) realize that data with serial number 1 is missing. The receiving end immediately sends ACKMSG(1) to inform the sender to resend the data with sequence number 1. When the retransmission data 1 is received, the receiving queue number 1 receives the identifier and changes to the reception completion status. When all three data are received, the receiving end uploads the three data of the receiving queue together to the application layer, and returns an ACKMSG (0) to notify the transmitting end that the receiving is completed; the sending queue changes the data sending status and deletes the data, waiting for the next one. Data is written to be sent. By using the high-power digital transmission and transmission machine and the RTCP transmission protocol, the data can be transmitted to the ground receiving end in real time.

7.2.5 Ground Control Platform Data Transmission After the digital transmission picture receives the packaged data and image data sent by the host, the packaged data is transmitted to the ground control platform computer via RS232 to USB for LZMA data decompression, and the required information is extracted according to the agreed protocol. Real-time display of spectral line data in the analysis software achieves digital stability, nuclear identification, generation of trajectory maps and radioactive GIS [10] maps, and other functions. In addition, the image data is transmitted from the digital transmission to the HDMI data acquisition card for data analysis, and the acquisition card transmits the parsed image data to the ground control platform computer, and displays the high-altitude image information in real time in the spectrum analysis software.

7 Real-Time Data Transmission Design of Unmanned Aerial …

73

7.3 System Test The experiment aims to test whether the data transmission system designed in this paper can realize the synchronous transmission of data and images under the condition that the ground receiving end is unobstructed at 40 km and continuously works for 5 h. The γ energy spectrum and the image of the HD camera are displayed in real time on the ground. The test result display is shown in Fig. 7.3. The whole system is shown in Fig. 7.4.

Fig. 7.3 Test result display

Fig. 7.4 System physical map

74

X. Zhong et al.

Table 7.2 Data packet loss retransmission test under 1 s measurement period Work time/h

Timeout retransmission/Times

Amount of data transferred/MB

Lost data volume/MB

0.5

3

14.23

0

1

5

28.56

0

2

9

57.12

0

3

13

85.62

0

5

20

142.85

0

Table 7.3 Storage data upload test after disconnected storage reconnect

Wire break storage/MB

Transmission time/秒

Lost data volume/MB

0.079

0.351

0

0.237

1.055

0

0.474

2.110

0

4.474

19.919

0

14.226

63.337

0

At the same time, the results of transmitting data stored after data timeout retransmission and disconnection recovery in the case of network anomaly are tested, as shown in Tables 7.2 and 7.3. It can be seen from the above diagram that the UAV gamma spectroscopy real-time data transmission system designed in this paper can transmit data and images in real time at 40 km, realizing real-time display of γ energy spectrum and camera images. In addition, in the case of abnormal network communication, data transmission can be resumed in time by means of retransmission and offline storage to ensure that data is not lost.

7.4 Conclusion The UAV gamma spectroscopy system uses four large volume NaI(TL) scintillator detectors to improve the instantaneous counting rate. The GPS second pulse is used as the time reference for each data information acquisition, which ensures time synchronization and improves data accuracy. In the industrial computer program, the LZMA algorithm is used to greatly compress the data to improve the transmission efficiency. At the same time, the TCP/IP and RTCP transmission protocols are used to transmit the data and image information to the ground control platform for spectral line shaping and real-time image display. It allows the operator to perform real-time radiological assessment of the detection area. The reliability and real-time performance

7 Real-Time Data Transmission Design of Unmanned Aerial …

75

of data transmission in unmanned aerial vehicle gamma spectroscopy measurement systems plays a crucial role in the measurement results of environmental radiation monitoring. Acknowledgments This work is supported by the National Natural Science Foundation of China, project number 41304117.

References 1. Liu, Y., Zhang, Z., Liu, Q.: Overview of Aeronautical Gamma spectroscopy measurement in China. Uranium Min. Metall. 26, 80–82 (2007) 2. Hua, Y., Ge, L., Luo, Y. et al.: Research on feature extraction information of aeronautical gamma spectroscopy based on data fusion. Nuclear Tech. 36, 2013 (2013) 3. Chen, L., Wei, Y., Qu, J.: Nuclide recognition algorithm in portable gamma spectrometer. J. Tsinghua Univ.(Science and Technology) 49, 635–638 (2009) 4. Yong, Wu: Real-time correction algorithm of reference source based on GPS second pulse. Infrared Laser Eng 43, 139–144 (2014) 5. Luo, B., Fei, X.: Research and application of multi-threading technology. J. Comput. Res. Dev. 37, 409–412 (2000) 6. Jian, Liu, Shengle, Li, Ziying, Wang, et al.: Research on application of compressed storage based on LZMA in database. J. Geod. Geodyn. 29, 144–147 (2009) 7. Chase, J.: High Definition Multimedia Interface, pp. 481–490. Springer, Berlin, Heidelberg (2012) 8. Perez, H., Gutierrez, J.J.: Modeling the QoS parameters of DDS for event-driven real-time applications. J. Syst. Soft. 104, 126–140 (2015) 9. Chun, L., Dajie, L.: Discussion on the application and research hotspot of GIS. Mod. Surv. Mapp. 26, 8–10 (2003)

Chapter 8

0.8 GHz Low-Noise Amplifier Design Xueshi Hou, Xiaoling Zhong, Zhilong Zhao, Liangyi Deng, Han Mei, Xue Wei, Yuting Jiang, Baiqiu Liu, and Yong Fang

Abstract In a receiving system such as a base station, the receiver front end needs to place a low-noise amplifier (LNA), and its performance will directly affect the performance of the entire receiver. Generally, the signal received by the antenna is weak. The function of the low-noise amplifier is to amplify the received useful signal and reduce the noise interference, so that the system can demodulate the required information data and improve the sensitivity of the receiving system. In this paper, the low-noise amplifier matching network is taken as the research direction, and two related methods are proposed. (1) In the matching design, the method of mixing and matching the lumped elements and microstrip lines on the Smith chart is adopted to design a reasonable matching circuit, which reduces the complexity of the subsequent overall optimization process. (2) By adding the equal standing wave ratio ring on the basis of the conventional equal noise circle and the equal gain circle, it is possible to more intuitively select an appropriate impedance matching point in the same Smith chart. According to the above method, a narrowband low-noise amplifier operating at 0.8 GHz is designed.

8.1 Introduction The emergence of gallium arsenide FET amplifiers in the 1960s has greatly increased the operating frequency of amplifiers. In the late 1980s, advances in process technology led to the development of new semiconductor components such as high electron mobility transistors (HEMTs) and heterojunction bipolar transistors (HBT) [1–4]. These two semiconductor components have low-noise characteristics, which are more suitable for low-noise scenarios compared to GaAs FETs [5–8]. In the 1980s, a French company called Thomson-CSF produced the first low-noise, high-electron mobility transistor. The HEMT operating frequency band is lower X. Hou (B) · X. Zhong · Z. Zhao · L. Deng · H. Mei · X. Wei · Y. Jiang · B. Liu · Y. Fang College of Information Science and Technology, Chengdu University of Technology, 1#, Dongsanlu, Erxianqiao, Chengdu 610059, Sichuan, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_8

77

78

X. Hou et al.

than 10 GHz, the gain is 10.3 dB, and the noise figure is 2.3 dB [9, 10]. First select the appropriate crystal amplifier tube, and then select the appropriate static working point through DC simulation to design the bias circuit and perform stability analysis. The input and output are, respectively, L-type and T-type matching networks. The circuit diagram and simulation results of the T-matching network are obtained through the ADS software, and finally the joint simulation optimization design of the whole circuit is performed.

8.2 Low-Noise Amplifier Design Flow In general, designing a low-noise amplifier first requires selecting a suitable crystal amplifier tube and a suitable quiescent operating point according to the design specifications, performing bias circuit design and stability analysis through DC simulation, and then analyzing according to gain and noise figure. The input–output matching design is required. After obtaining the amplifier circuit diagram and simulation results, the overall circuit should be jointly simulated and optimized, finally processed into physical objects and tested and debugged. The design process is shown in Fig. 8.1. Before designing the bias circuit, select the appropriate static operating point based on noise, gain, linearity, and other metrics. Looking at the worksheet of the ATF35143 chip, the relationship between the drain-source current and noise figure and the gain at 900 MHz operating frequency can be seen in Fig. 8.2. It can be observed from Fig. 8.2 that as the drain-source current increases, the noise figure of the ATF35143 gradually decreases, and when it exceeds 30 mA, the noise figure starts to increase. The trend of gain is opposite to the noise figure. Noise performance and gain performance are best when the drain-source current is between 20 and 40 mA. Then use ADS to perform DC simulation analysis on the low-noise amplifier chip, and the relationship between them can be obtained as shown in Fig. 8.3. As shown, it can be seen that when Vds = 3 V, Ids = 30 mA, the gate-source voltage VGS = −0.4 V. After determining the values of the gate-source voltage

Fig. 8.1 Low-noise amplifier design flowchart

8 0.8 GHz Low-Noise Amplifier Design

79

Fig. 8.2 Ids and noise figure and gain relationship

Fig. 8.3 Static working point simulation and Vds fixed, VGS and Ids relationship

and the drain-source voltage, the design of the bias circuit can be performed. Since the gate-source voltage of the transistor is negative and the drain-source voltage is positive at this quiescent operating point, considering the operability of the amplifier’s debugging process, the design decided to use dual-power supplies. Its bias circuit is shown in Fig. 8.4. In the simulation circuit, C1 and C2 are DC blocking capacitors, and the DC signal directly affects the input and output. L1 and L2 are inductors, which are equivalent to RF chokes. C3, C4, C5, C6, C7, and C8 are RF grounded capacitors whose capacitance units are PF, NF, and UF, respectively. When the high-frequency

80

X. Hou et al.

Fig. 8.4 DC bias circuit diagram of ATF35143

power supply ripple is filtered out, they are equivalent to a short circuit condition, and the RF signal of the leakage choke inductor can be transmitted to the ground. The circuit simulation is shown in Fig. 8.4.

8.3 Research and Design of Matching Networks 8.3.1 Selection of Impedance Points This work task is to design a low-noise amplifier. The minimum noise matching should be chosen. However, this design is a single-stage low-noise amplifier. If the minimum noise is matched, the gain will be slightly lower and the design requirements will not be met. Therefore, it is necessary to balance the gain and noise. A narrowband low-noise amplifier with a small noise figure and high gain is designed by combining an equal noise figure circle with an equal gain circle. After debugging the equal gain circle and the equal noise figure circle at 0.8 GHz, the input impedance point is finally selected as s = (36.648 + j128.073), and the output impedance point is L = (22.251 + j17.089). Its equal noise figure circle and equal gain circle are shown in Fig. 8.5.

8 0.8 GHz Low-Noise Amplifier Design

81

Fig. 8.5 Combine the equal noise figure circle with the equal gain circle to select the source impedance point

8.3.2 Matching Network Design The matching method proposed in this paper is that the lumped parameters and the microstrip line are mixed and matched on the Smith chart. When the matching is carried out, the influence of the microstrip line between the components is taken into consideration. This method can quickly design a reasonable matching circuit. The matching process is shown in Fig. 8.6a, b. After the matching, the lumped component value and the size of the microstrip line need to be added to the amplifying circuit. Note here that the lumped component should be the actual component. The length and width of the microstrip line can be calculated by the LineCalc tool of the ADS software. When matching, pay attention to the Deg value of the microstrip line. If the Deg value is too large, the final calculated microstrip line size will become very large. In the LineCalc calculation interface, the FR-4 plate parameters should be adjusted first. The calculation method of the microstrip parameters is shown in Fig. 8.7. The schematic diagram obtained after joining the matching network is shown in Fig. 8.8. Run the simulation after ADS software debugging optimization. The simulation results are shown in Fig. 8.9. From the simulation results, it can be concluded that the amplifier is stable around 0.8 GHz, the noise figure is less than 0.5 dB, the gain is greater than 18 dB, and the input reflection coefficient S11 and the output reflection coefficient S22 are both less than -10 dB.

82

X. Hou et al.

Fig. 8.6 a Input matching design after adding microstrip line. b Output matching design after adding microstrip line

8.3.3 LNA Production Debugging and Analysis According to the circuit simulation of Fig. 8.8, the circuit board shown in the following figure is fabricated, and the capacitor inductor, the amplifier chip, and the SMA are soldered, and the soldered circuit is as shown in Fig. 8.10.

8 0.8 GHz Low-Noise Amplifier Design

83

Fig. 8.7 Calculation method of length and width of microstrip line

Fig. 8.8 Low-noise amplifier schematic

8.3.4 Circuit Debugging and Analysis During commissioning, check the weld quality, test the voltage at each quiescent operating point, check that the transistor voltage at each quiescent operating point

84

X. Hou et al.

(a)

(b)

Fig. 8.9 a Simulation noise and gain. b Input and output reflection coefficient Fig. 8.10 Circuit diagram

8 0.8 GHz Low-Noise Amplifier Design

85

Fig. 8.11 Dual-supply voltage and transistor Ids current value

Fig. 8.12 Actual test input and output reflection coefficients

is normal, that the bias is consistent with the simulation design, and that the bias current is close to 30 mA. Dual-supply voltage and transistor Ids current values are shown in Fig. 8.11. The instrument used for S-parameter measurements in this topic is a vector network analyzer. When preparing to test a low-noise amplifier, first check that each port of the vector network analyzer is isolated from DC current, and then apply a DC voltage to adjust the output power of the corresponding network analyzer. After checking that the noise amplifier is properly connected to the vector network analyzer, start testing each result. Figures 8.12 and 8.13 show the actual test parameter values for the LNA input and output reflection coefficients and gain, respectively. It can be seen from the test results that at a frequency of about 800 MHz, S11 can reach about −14.2 dB, S22 can reach 16.7 dB, and the gain can reach 15.4 dB, which basically meets the design requirements. Compared to the simulation results, the gain difference is approximately 3 dB and the center frequency offset is approximately 10 MHz. Physical measurements have a small amount of error compared to the simulation results. The possible reasons are as follows: (1) Soldering with SMA interface and components installed at the input and output of the amplifier may cause some loss; (2) The value of the actual circuit component has a certain error compared with the value of the component during the simulation; (3) The FR4 dielectric substrate material and process have some influence on the designed low-noise amplifier.

86

X. Hou et al.

Fig. 8.13 The actual test parameter value of the LNA gain

8.4 Conclusions This paper details the design flow of the low-noise amplifier and the bias circuit design and static operating point stability analysis. A 0.8 GHz low-noise amplifier is designed through the input–output matching network, and the gain is greater than 14 dB in the 0.75–0.85 GHz band. Acknowledgments This work is supported by the National Natural Science Foundation of China, project number 41574137.

References 1. Hong, Y.H., Yoo, H.: A low Flicker noise RF front-end with resonating inductor for 2.4 GHz ISM band. IEEE International Workshop on Radio-Frequency Integration Technology, pp. 222–226. Rasa Sentosa Resort (2007) 2. Alzaher, H.A., Ismail, M.: A CMOS fully balanced differential difference amplifier and its applications. IEEE Transactions on Circuits and Systems II: Analog and Digital Signal Processing, vol. 48, no. 6, pp. 614–620 (2001) 3. Ivanov, B.I., Grajcar, M., Novikov, I.L.: A microwave cryogenic low-noise amplifier based on sige heterostructures. Tech Phys Lett 42(4), 380–383 (2016) 4. Karimlou, A., Jafarnejad, R., Sobhi, J.: An inductor-less Sub-mW low noise amplifier for wireless sensor network applications. Integr. Vlsi J. 52, 316–322 (2016) 5. Liu, R.C, Deng, K.L., Wang, H.: A 0.6-22-GHz broadband CMOS distributed amplifier. Radio Frequency Integrated Circuits. IEEE (2003) 6. Niclas, K.B.: The exact noise figure of amplifiers with parallel feedback and lossy matching circuits. IEEE Trans. Microw Theory Tech 30(5) (1982) 7. Sturm, O.E., Orton, R., Grindlay, J.: The Mammalian MAPK/ERK pathway exhibits properties of a negative feedback amplifier. Sci. Signal. 3(153), ra90 (2010)

8 0.8 GHz Low-Noise Amplifier Design

87

8. Chang, J. Aumentado, W.-T.W., Bardin, J.C.: Noise measurement of cryogenic low noise amplifiers using a tunnel-junction shot-noise source. IEEE MTT-S International Microwave Symposium (IMS), San Francisco, CA, pp. 1–4 (2016) 9. Shaeffer, D.K., Lee, T.H.: A 1.5-V, 1.5-GHz CMOS low noise amplifier. Symp. Vlsi Circuits Dig .32.5, 745–759 (1997) 10. Shaeffer, D.K., Lee, T.H.: A 1.5 V, 1.5 GHz CMOS low noise amplifier. Symposium on Vlsi Circuits, Digest of Technical Papers (1996)

Chapter 9

Optimization Analysis of Laying Length of Thin-Wall Micro-Spray Belt Under Different Slope Conditions Jin Yi Wang, Lu Hua Yang, Wan Li Gou, and Yan Hui Dong

Abstract To optimize the laying length of the thin-walled micro-spraying belt, the field layout way of the micro-spraying belt was considered to determine the water head distribution coefficient β as 0.55 in lateral pipe. The allowable water head deviation of the micro-spraying belt was calculated. Combined with the energy conservation law, the models of the relative pressure function between the head end and the tail end of the micro-spraying belt under different slope-laying conditions were established, so were the mathematical relationships between the allowable head deviation of the micro-spraying belt and, the head end and the tail end relative pressure. Thus, the optimization of the laying length and slope was analyzed. It is found that the relationship between the relative pressure and the laying length is a decreasing function at the adverse slope while a parabolic function at the down slope. When the laying length of the micro-spraying belt is arranged to L0 , the relative pressure P (L0 ) gets the maximum value. The paper put forward the laying length formula of the micro-spraying belt under different laying conditions—down slope, flat slope and adverse slope. The laying length of down slope of the micro-spraying belt with fold diameter of 43 and 64 mm was calculated under the condition of adverse slope at an incline of 0.01, 0.03 and 0.05 degrees and flat slope, respectively, aiming to provide a theoretical basis for engineering plan and design.

J. Y. Wang · L. H. Yang (B) Tianjin Agricultural University, Tianjin, China e-mail: [email protected] J. Y. Wang e-mail: [email protected] W. L. Gou Dayu Water-Saving Company, Tianjin, China e-mail: [email protected] Y. H. Dong Xi’an Technological University, Xi’an, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_9

89

90

J. Wang et al.

9.1 Introduction Compared with other irrigation equipment, micro-spraying belt has the characteristics of low investment, strong anti-blocking ability and easy installation. At present, some scholars have been studying the laying length and slope of micro-spraying belt and achievements were obtained. Wang et al. [1] studied the influence of the laying length of micro-spraying belt on its uniformity under the low-pressure head, and it is found that the uniformity of the micro-spraying belt increased with the increase of the pressure, while decreased with the increase of the laying length when the working pressure was constant. Li et al. [2] analyzed the influence of the pipe diameter, bore diameter and laying slope of micro-spraying belt of the typical model on the uniformity of water spray. Wang [3] designed the water-fertilizer micro-sprinkle-irrigation system, and conducted simulation analysis on the water spray uniformity of micro-spraying belt with AMESim software. Xi et al. [4] took built-in drip irrigation belt and thin-walled drip irrigation belt as research objects, and compared the uniformity of irrigation water application under four laying lengths of drip irrigation belt and the interaction of four inlet water pressures. Ma et al. [5] carried out experimental research on the influence of the laying length, pressure head and topographic gradient of drip irrigation belt for three factor on the effect law of the uniformity coefficient of drip irrigation belt. Wang et al. [6] tested the irrigation uniformity of drip irrigation belt in the field and found that when the laying length of drip irrigation belt was small, the irrigation uniformity of the tested products was very high. Sun [7] found that both the head pressure and the laying length of drip irrigation belt had a great impact on the uniformity of drip irrigation. When the laying length and the head pressure were constant, the overall water delivery capacity of drip irrigation belt did not change greatly, but the uniformity changed significantly and decreased with the increase of absolute slope value. Most of the above researches focus on irrigation devices such as drip irrigation belt or sprinkler head, and lack of optimization research on laying length and slope of multi-hole micro-spraying belt on slope. Therefore, from-micro spray belt laying length and slope, the paper carry out field experiment, build up optimization mathematical model to determine the formula of micro-spray belt length in the down slope, flat slope, adverse slope and provide guidance for agricultural irrigation.

9 Optimization Analysis of Laying Length …

91

9.2 Analysis on Allowable Laying Length of Micro-Spray Belt 9.2.1 Allowable Head Deviation in Lateral Pipe According to the reasonable distribution on branches and lateral pipes of the allowable pressure difference in the micro-irrigation design unit [8], and in combination with the laying way of micro-spray belt in the field, the water head coefficient of lateral pipe distribution is set as 0.55. According to Eq. (9.1), the allowable head deviation of micro-spray belt was calculated. (9.1) [h] = h v h d

(9.2)

  1−x 1 qv [h v ] = qv 1 + 0.15 x x

(9.3)

where β is the capillary head loss coefficient; [hv ] is head deviation rate; qv is hole of outlet flow rate (L/h); X is the flow pattern exponent; [h] is the allowable head deviation of irrigation district; Hd is head of orifice working pressure (m).

9.2.2 The Formula of the Head Loss Along the Micro-Spray Belt Refer to technical specifications for micro-irrigation engineering, considering that the porosity coefficient tends to be fixed when the plastic micro-spray belt is laid for a long, the loss of water head along the micro-spray belt can be calculated by the following Eq. (9.4): hf = K

Qm L db

(9.4)

where hf is head loss along the way (m); Q is pipe flow (L/h); d is pipe diameter (mm); L is pipe length (m); K is friction coefficient; m is flow index; b is pipe diameter index. In 2018, 12 different types of water head loss of plastic micro-spray belts were tested in Tianjin agricultural water-saving engineering center. According to the test results, the standard recommendation for comparison analysis is comprehensive determination of formula parameters. Next, K is 0.038, m is 1.653, b is 4.168. The

92

J. Wang et al.

formula is applicable to the calculation of the head loss along the thin-wall microspray belt with folded diameter of N43–N64 mm (pipe diameter of 27–40 mm), pressure head of 1–8 m and laying length of 20–100 m.

9.2.3 Pressure of Micro-Spray Belt at Any Point Under Different Slopes According to the variation of the slopes I of the micro-spray belt, the formula for calculating the pressure at any point of the micro-spray belt is determined as shown in Eq. (9.5). H (L) = H1 +0.98 sin(arctan(I ))L − h f (L)

(9.5)

where H(L) is the pressure at any point of L (m); H1 is head end pressure (m); I is slope; L is the distance from the first hole to any point (m); Hf (L) is the water head loss along the length of L (m). First, on the basis of q = CHx , with q/S0 of each point is the average flow when q is distributed on S0. Combined with the head loss along the way, the head loss along the way is calculated by the outlet flow of hole, as shown in Eq. (9.6). L dh f (L) = K h f (L) = K

1 db



nq s0

m

1 db



nq S0

m (N S 0 −L)m dl

0

 1  (N S 0 )m+1 − (N S 0 −L)m+1 m+1

(9.6)

where K is the resistance coefficient along the way of micro-spray belt; m is the flow index; b is pipe diameter index; d is pipe diameter, (mm); q is hole flow (m3 /s); S0 is spacing of hole groups, (m); N is the number of hole groups; n is the number of holes in the hole group.

9.3 Relative Pressure at Any Point Under Different Slopes In order to analyze the laying length of micro-spray belt under different slopes, relative pressure is introduced, that is, the relative pressure is the difference between the pressure of any hole on the micro-spray belt and the pressure at the head end of the micro-spray belt, as shown in Eq. (9.7).

9 Optimization Analysis of Laying Length …

93

P(L) = H(L) − H1

(9.7)

where P(L) represents the relative pressure (m). Take Eqs. (9.5) and (9.6) into Eq. (9.7) to get Eq. (9.8). K P(L) = 0.98 sin(arctan(I ))L − b d



nq S0

m

 1  (N S 0 )m+1 − (N S 0 −L)m+1 m+1 (9.8)

At the end of the pipeline, the distance between the end hole and the head hole is expressed by Le . According to the hole group spacing of S0 , the relative pressure at the end of micro-spray belt can be obtained, as shown in Eq. (9.8). L e = (N − 1)S0

(9.9)

where Le is the distance between the hole at the head end and the hole at the end of the micro-spray belt (m); N is the total number of holes in the micro-spray belt; S0 is the hole group spacing of micro-spray belt (m). Take the deformation of Eq. (9.9) into Eq. (9.8) to get Eq. (9.10). P( L e ) = 0.98 sin(arctan(I )) L e −

 K  nq m 1  ( L e +S0)m+1 − (S0)m+1 b m+1 d S0 (9.10)

where all parameters have the same meaning as described earlier. The function P(Le) is a convex function. The function graph is shown in Fig. 9.1. Fig. 9.1 Image schematic diagram of the approximate function of relative pressure and laying length of thin wall micro spray belt on downhill, flat and adverse slopes

94

J. Wang et al.

9.4 The Maximum Laying Length Under Different Slopes 9.4.1 Determination of the Laying Length of Micro-Spray Belt During Flat Slope When laying micro-spray belt on flat slope, the pressure of micro-spray belt decreases with the increase of laying length, that is, the pressure on the head end hole is the maximum and the pressure on the end hole is the minimum. The relation between the relative pressure on the head end and the allowable pressure is shown in Eq. (9.11). (9.11) According to Eq. (9.11), combined with Eq. (9.10), it can be obtained that, when laying flat slope the maximum laying length of micro-spray belt Lm is shown in Eq. (9.12). (9.12)

where all parameters have the same meaning as described earlier.

9.4.2 Determination of Laying Length of Micro-Spray Belt in Adverse Slope In adverse slope condition, the laying length of micro-spray belt is shorter than that of flat slope, because in adverse slope condition, the laying slope of micro-spray belt is negative as well as topographic. By combining Eqs. (9.10) and (9.11), the expression of the relationship between the laying length of micro-spray belt in adverse slope can be obtained, as shown in Eq. (9.13).

(9.13) where all parameters have the same meaning as described earlier.

9 Optimization Analysis of Laying Length …

95

9.4.3 Determination of Laying Length of Micro-Spray Belt in Down Slope When the micro-spray belt is laid with down slope, the laying length is longer than that in the flat slope under the suitable slope (not exceeding the maximum slope of Ia allowed by the micro-spray strip). When the laying slope of micro-spray belt is too large, the pressure difference between the head and end of micro-spray belt will be too large under the action of gravity, which will affect the uniformity of water output of micro-spray belt. Therefore, when the laying slope of micro-spraying belt does not exceed the maximum allowable laying slope with Ia , the calculation formula of laying length of micro-spraying belt with down slope can be obtained by combining the water head deviation for down slope and Eq. (9.10), as shown in Eq. (9.14).

(9.14) where all parameters have the same meaning as described earlier.

9.5 Laying Length Analysis Select N43 and N64 mm micro-spraying belt commonly used in the market to respectively calculate the water head deviation of micro-spraying belt and the laying length of down slope, flat slope and adverse slope according to the above deduced formula, and analyze the laying length rule of micro-spraying belt under different slopes. The resistance coefficient K, flow index m and pipe diameter index b obtained through multiple linear regression calculation of water head loss parameters of thin-wall micro-spray belt by Gou et al. [9], and the values, respectively, are 0.038, 1.653 and 4.168.

9.5.1 Calculation of Head Deviation and Deviation Rate According to formulas (9.1) and (9.2), the deviation rate and head deviation with water head of 43 and 64 mm micro-spraying belt are calculated, respectively. The calculated results are shown in Table 9.1.

96

J. Wang et al.

Table 9.1 Water head deviation and deviation rate of thin-wall micro-spray belt Fold diameter (mm)

Flow coefficient C

Flow index X

Water head deviation rate [hv]

Water head deviation [h]

N43

3.785

0.537

0.382

2.675

N64

3.398

0.546

0.375

3.754

Table 9.2 Calculating table of laying length (m) of micro-spray belt for different slope Fold diameter (mm)

Gradient of down slope (I)

Flat slope

Gradient of adverse slope (I)

0.01

0.03

0.05

0

0.01

0.03

0.05

N43

31.73

34.4

37.1

30.42

29.13

26.61

24.23

N64

76.99

87.67

98.41

71.76

66.67

57.05

48.51

9.5.2 Calculation of Laying Length for Different Slope The ground slope of the down slope was set at 0.01, 0.03 and 0.05, respectively. Combined with formula (9.14), the laying length of the micro-spray belt for down slope was calculated. The calculation results are shown in Table 9.2. Through analysis, it is found that the laying slope of micro-spray belt is the same, and with the fold diameter gets larger, the laying length gets larger too. When the same folding diameter micro-spray belt is laid at different slope degrees, when the slope gets larger, the laying length gets longer. According to formula (9.12), the laying length of micro-spray belt in flat slope is calculated, respectively, and the calculation results are shown in Table 9.2. The ground slope of the adverse slope was set at 0.01, 0.03 and 0.05, respectively, and the laying length of the adverse slope of the micro-spray belt was calculated by combining formula (9.13). The calculation results are shown in Table 9.2.

9.6 Conclusions In this paper, the commonly used model of thin-walled micro-spray belt and experimental study on the hydraulic performance of the irrigation district water head deviation as constraints, based on the analysis of the head end and the tail end relative pressure, allows the water head deviation. Besides, combined with the water head loss along the hydraulic performance parameters of the mathematical model, we thus obtained in down slope, flat slope, the adverse slope conditions for micro-spray belt length calculation formula, and calculated the fold diameter as 43 and 64 mm of micro-spray belt in the down slope, flat slope, the adverse slope for laying length.

9 Optimization Analysis of Laying Length …

97

Acknowledgments Thanks for the support of Natural Science Basic Research Program Fund (2019JQ-838) in ShanXi province and Haihe Water Resources Committee Fund (YH002101) of the Ministry of Water Resources, and thanks for the guidance and help provided by teacher of LuHua Yang and senior brothers and sisters in this paper.

References 1. Wang, Y.B., Li, D.X.: Study on spray uniformity of low head micro-spray belt. People Yellow River l 146–149 (2018) 2. Li, Y., Hao, M., Wang, K., Fan, J.: Modeling and experimental research of field water and fertilizer integrated micro-sprinkler irrigation system based on AMESim. Chin. J. Agric. Mech. 38, 53–60 (2017) 3. Wang, Y.C.: Design and simulation of integrated micro-sprinkler irrigation system for watersaving irrigation. Shanxi Prov. Water Conserv. 98–100 (2017) 4. Xi, Q.L., Ge, G.F., Zhou, F.: Response of irrigation uniformity of two drip irrigation belts to laying length and inlet pressure. J. Irrig. Drain. 37, 78–83 (2018) 5. Ma, X.P., Gong, S.H., Wang, J.D., Yu, Y.D., Dong, Y.F.: Experimental study on irrigation uniformity coefficient of drip irrigation belt under low pressure. J. Irrig. Drain. 29, 6–10 (2010) 6. Wang, Y., Fan, Z.L., Feng, H.X., Chen, M.Y.: Field test and analysis of irrigation uniformity of drip irrigation belt. J. Gansu Agric. Univ. 35, 66–69 (2000) 7. Sun, C.: Experiment on flow uniformity of drip irrigation belt and analysis of drip irrigation system in Mountain Terraces. Shanxi Agric. Univ. 14–18 (2016) 8. Keller, J., Karmeli, D.: Trickle irrigation design parameters. Trans. ASAE 17, 678–0684 (1974) 9. Gou, W.L., Yang, L.H., Di, Z.G.: Experimental study on water head loss along thin-wall sprayirrigation pipe. J. Irrig. Drain. 38, 79–83 (2019)

Chapter 10

L-Band Ultra-Wideband Low-Noise Amplifier Design Xiaoling Zhong, Haoxuan Sheng, Yong Fang, Yong Guo, Baiqiu Liu, Zhilong Zhao, and Yangyang Wang

Abstract The low-noise amplifier (LNA) is located at the front end of the RF receiver and is the core component of the microwave receiver. Its performance directly affects the sensitivity of the receiver and the noise figure of the entire system. At the same time, ultra-wideband technology has developed rapidly, and ultra-wideband lownoise amplifiers with excellent performance have high research value. The main content of this paper is to design and implement an L-band ultra-wideband lownoise amplifier. The wideband amplifier designed in this paper introduces a parallel negative feedback branch, which is powered by a single-supply mode, using a common source unipolar amplifier structure. By optimizing the size of each device and microstrip line, a broadband amplifier with matching network design frequency of 0.65–1.35 GHz, gain G > 9 dB, gain flatness is uth ,the new diet data is porcine normal diet data. If u(x new ) < uth , the new diet data is porcine abnormal diet data. In this paper, uth = 0.6. The volume of hypersphere is mainly controlled by the penalty parameter C and kernel parameter σ. With the increase in C, the punishment on the samples exclude from the hypersphere becomes heavier [11]. The boundary of SVDD is sensitive to the kernel parameter σ. With the increase of the σ, the boundary undergoes a complex change [12]. The selection of the two parameters has a great influence on the performance of SVDD. The PSO is employed to optimize these two parameters in this paper.

142

S. Zhang et al.

14.3 The Parameter Optimization of SVDD Based on PSO Traditional PSO searches for the optimal solution through cooperation and information sharing among the particles in the swarm. In the n-dimensional space, there are m particles in the swarm, which is represented as X = [x 1 , x 2 , …, x m ]. The position and velocity of the ith particle are represented as x i = [x i1 , x i2 , …, x in ] and vi = [vi1 , vi2 , …, vin ]. The best previous position of the ith particle is recorded as Z best ,i = [zi1 , zi2 , …, zin ]. The best position of the swarm is denoted as Z g,best = [zg1 , zg2 , …, zgn ]. The velocities and positions of the particles in the dth dimension are updated as the following equations [13]: k+1 k k k = wvid + c1r1 (Zkbest,id − xid ) + c2 r2 × (Zkg,best,d − xid ) vid k+1 k+1 k xid = xid + vid

(14.8) (14.9)

where d = 1,2,…,n; w is the inertia factor; k is the number of iterations; c1 , c2 are acceleration factors; r 1 , r 2 are two random numbers within [0,1]. In order to avoid the local minimum problems, particle mutation is adopted to improve PSO in the optimization process of SVDD parameters. In the iterative process, the particles mutate with the probability p(k). A large mutation probability is set at the beginning of iteration to avoid premature local minimum. As the number of iterations increases, the mutation probability decreases gradually to avoid the mutation probability of particles is too high to converge. The strategies of mutation are as follows:  (xmax − xmin ) × rand() + xmin , rand() < p(k) k (14.10) xi = , rand() ≥ p(k) xik−1 where x ki is the ith particle of kth iteration; x max is the maximum position of the particle; x min is the minimum position of the particle; rand() is a random number which lies in [0,1]; p(k) is the mutation probability, which is calculated as stated below: p(k) = pmin + ( pmax − pmin )(N − k)/N

(14.11)

where pmin is the minimum mutation probability; pmax is the maximum mutation probability; N is the maximum number of iterations.

14 The Research on the Judgement Method for Porcine Abnormal Diet …

143

14.4 Experimental Results and Analysis In this study, experimental data were collected from a large-scale pig farm located in Shanxi Province, China. The foraging style of pigs is free food intake. In order to collect porcine diet data, a collection device of porcine diet data based on ultrasonic has been designed. It can collect daily diet times and diet time in the pigpen. The collection device is shown in Fig. 14.1. In our experiment, 1000 normal diet data are used as training data, 150 abnormal diet data, and 150 normal diet data are used as test data.

14.4.1 The Comparison Results of Decision Function Before and After Improvement In order to evaluate the judgment results quantitatively, precision, recall, and accuracy are adopted as evaluation indicators. The calculation formulas of the three indicators are as follows [14]:

A=

P=

TP T P + FP

(14.12)

R=

TP T P + FN

(14.13)

TP +TN T P + FP + FN + T N

(14.14)

where P is the precision; R is the recall; A is the accuracy; TP is denoted as the number of correctly judged normal data; FN is denoted as the number of normal data that are Fig. 14.1 The collection device of porcine diet data based on ultrasonic

144

S. Zhang et al.

Table 14.1 The judgment results of the decision function before and after the improvement Parameters C = 0.18 σ = 0.62

Initial decision function (%)

Improved decision function (%)

P

P

100

R

A

R

A

62.67

81.33

99.19

81.33

90.33

C = 0.45 σ = 2.55

95.24

53.33

75.33

96.97

85.33

91.33

C = 0.72 σ = 3.66

95.40

55.33

76.33

96.88

82.67

90

C = 0.95 σ = 7.55

89.83

70.67

81.33

90.08

78.67

85

falsely judged as abnormal data; FP is denoted as the number of abnormal data that are falsely judged as normal data; TN is denoted as the number of correctly judged abnormal data. The judgment results of the decision function before and after the improvement are shown in Table 14.1. Table 14.1 shows that the judgment results with different parameters are different. The results of improved decision function with different parameters show less volatility than the initial decision function. The improved method is more stable for parameter variation. Under different parameters, the precision of the improved decision function is not much different from that of the initial decision function, but the recall and accuracy are better than the initial decision function. This is because the improved decision function has better judgment ability for the data at the edge of the hypersphere. It can effectively reduce the misjudgment of normal diet data and FN. Therefore, recall and accuracy are improved. Because the parameters have an influence on the judgment results of SVDD and the trial and error method is difficult to determine the optimal parameters, the improved PSO is used to optimize the parameters of SVDD.

14.4.2 The Judgment Results for Porcine Abnormal Diet Data Based on Improved PSO-SVDD The improved PSO is used to optimize the parameters of SVDD. The maximum number of iterations is 200 and the number of particles is 60. The optimized parameters are used to construct the judgment model for porcine abnormal diet. And the test data are judged by the model. The experimental results are shown in Table 14.2. Table 14.2 The judgment results after parameters optimization The optimized parameters

C = 0.63, σ = 4.88

Initial decision function (%)

Improved decision function (%)

P

R

A

P

R

A

95.38

82.67

89.33

95.83

92

94

14 The Research on the Judgement Method for Porcine Abnormal Diet …

145

Fig. 14.2 The comparison diagram of judgment results

Table 14.2 shows that the optimized parameters of SVDD are C = 0.63,σ = 4.88. After the improvement of the decision function, the precision, recall, and accuracy reached 95.83%, 92%, and 94%, respectively. In order to display the results more intuitively, the judgment results before and after the improvement of decision function are shown in Fig. 14.2. Figure 14.2 intuitively shows that the improved PSO-SVDD can improve the performance of judgment for porcine abnormal diet. The effectiveness of the improved method is verified.

14.5 Conclusions In order to monitor the porcine abnormal diet, the judgment method for porcine abnormal diet based on improved PSO-SVDD is proposed. Because the penalty factor and kernel parameter of SVDD are difficult to determine, PSO is used to optimize the parameters of SVDD. To avoid the local minimum problems, particle mutation is adopted to improve PSO. Because of the fuzzification of the judgment for porcine abnormal diet, a fuzzy decision function is proposed in this paper. The experimental results show that the improved PSO-SVDD can improve the performance of judgment for porcine abnormal diet. It provides an effective method for the judgment of porcine abnormal diet through the diet data. Acknowledgements This study was supported by the National High Technology Research and Development Program of China (863 Program) (2013AA102306).

146

S. Zhang et al.

References 1. Brownbrandl, T.M., Rohrer, G.A.: Analysis of feeding behavior of group housed growingfinishing pigs. Comput. Electron. Agric. 99, 209–217 (2013) 2. Matthews, S.G., Miller, A.L., Clapp, J., et al.: Early detection of health and welfare compromises through automated detection of behavioural changes in pigs. Vet. J. 217, 43–51 (2016) 3. Maselyne, J., Saeys, W., Briene, P., et al.: Methods to construct feeding visits from RFID registrations of growing-finishing pigs at the feed trough. Comput. Electron. Agric. 128, 9–19 (2016) 4. Lind, N.M., Vinther, M., Hemmingsen, R.P., et al.: Validation of a digital video tracking system for recording pig locomotor behaviour. J. Neurosci. Methods 143, 123–132 (2005) 5. Huang, Y.S., Huang, Y.P., Wang, J.S., et al.: Quantification of pre-parturition restlessness in crated sows using ultrasonic measurement. Conf Proc IEEE Eng. Med. Biol. Soc. 18, 2446–2449 (2005) 6. Zhao, Y., Wang, S., Xiao, F.: Pattern recognition-based chillers fault detection method using support vector data description (SVDD). Appl. Energy 112, 1041–1048 (2013) 7. Wan, S., Chen, L., Dou, L., et al.: Mechanical fault diagnosis of HVCBs based on multi-feature entropy fusion and hybrid classifier. Entropy 20, 1–19 (2018) 8. Liu, J., Liu, J., Chen, H., et al.: Abnormal Energy identification of variable refrigerant flow air-conditioning systems based on data mining techniques. Appl. Therm. Eng. 150, 398–411 (2019) 9. Wenliao, D., Zhiqiang, G., Liangwen, W. et al.: Intelligent fault diagnosis of plunger pump in truck crane based on a hybrid fault diagnosis scheme. Intell. Control Autom. IEEE, 5361–5365 (2015) 10. Hamidzadeh, J., Namaei, N.: Belief-based Chaotic Algorithm for Support Vector Data Description. Soft. Comput. 23, 4289–4314 (2019) 11. Zhenchong, Z., Xiaodan, W.: Cost-sensitive SVDD models based on a sample selection approach. Appl. Intell. 48, 4247–4266 (2018) 12. Tan, C., Chen, H., Lin, Z., et al.: Category identification of textile fibers based on near-infrared spectroscopy combined with data description algorithms. Vib. Spectrosc. 100, 71–78 (2019) 13. Janani, R., Vijayarani, S.: Text document clustering using spectral clustering algorithm with particle swarm optimization. Expert Syst. Appl. 134, 192–200 (2019) 14. Pereira, S., Pinto, A., Alves, V., et al.: Brain tumor segmentation using convolutional neural networks in MRI images. IEEE Trans. Med. Imaging 35, 1 (2016)

Chapter 15

The Design of Two-Channel Outputs Switching Mode Power Supply Based on TOP100Y Cao Yong, Zheng Yifan, Wang Xiao, Zhou Heng, and Liu Yi

Abstract Because of heat energy consumption in the linear voltage-stabilized power supply, a two-channel output of +12 V and +5 V switching mode power supplies based on TOP100Y is designed, the structure and working mode of circuit system are analyzed briefly, and the design method of the high-frequency transformer parameter is introduced. Then the circuit capability is validated by analyzing the experimental result, the project of improving and optimizing is put forward, and the method of improving efficiency is explained in this paper.

15.1 Introduction The semi-active suspension system of a tracked armored vehicle controls the magnitude of vibration reduction by using the solidification degree of magneto rheological fluid in the shock absorber. When designing the program-controlled DC/DC power supply converter of the MRF control circuit system, the selected main control chip UC3843 and operational amplifier LM358 all need 12 V voltage, and ATMEGA128 MCU needs 5 V voltage. In addition, according to the specifications of the circuit system, the current of 100 mA can meet the requirements. The tracked armored C. Yong · Z. Yifan (B) · W. Xiao · L. Yi Army Academy of Armored Forces, Beijing, China e-mail: [email protected] C. Yong e-mail: [email protected] W. Xiao e-mail: [email protected] L. Yi e-mail: [email protected] Z. Heng Xinjiang Military Region, Xinjiang 830000, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_15

147

148

C. Yong et al.

vehicle can only provide about 24 V. Therefore, we need to design a dual auxiliary voltage regulator with input of +24 V, output of +12 V, and output of +5 V. Many kinds of circuits can meet this requirement. Here, to make the power supply meet the requirement of miniaturization, we should choose a circuit which is relatively simple and convenient to realize.

15.2 Scheme Selection 15.2.1 Basic Regulator Regulating Circuit The basic regulator circuit is the basic linear regulated power supply. According to the design requirements of the auxiliary power supply, the input voltage is +24 V, and the output voltages are +12 V and +5 V. Therefore, the basic regulator circuit needs to design two stages, the first stage of input is +24 V voltage, after adjusting the voltage stabilizer to +12 V voltage output, at the same time +12 V voltage as the input voltage of the next stage, and then after adjusting the voltage stabilizer to +5 V output. The design circuit is shown in Fig. 15.1. For the above linear regulated power supply circuit, in the first-stage circuit, the voltage drop of input and output is 24 V−12 V = 12 V. If a chip needs about 100 mA of current, the circuit will consume about 1.2 W of energy on the transistor Q1 in the form of heat. However, the transistor Q1 can’t withstand such a large amount of heat, which will easily cause its burnout. Similarly, the transistor Q2 in the secondstage circuit will withstand about 0.07 W of heat, which is not a problem. If this 0.07 W heat is added to Q1, then Q1 will burn out more easily. Therefore, the basic regulator volt-age regulator circuit as an auxiliary power supply can’t meet the system requirements [1]. Fig. 15.1 Linear regulated power supply circuit

Q1

+24V

+12V 10K

10K Q2

D1 12V

D2 5V

+5V

15 The Design of Two-Channel Outputs Switching Mode …

149

15.2.2 Circuit TOP Power Supply TOP power supply is widely used at present. Its main chip is TOP Switch series integrated chip of high-frequency switching power supply, which is developed by Power Integrations Inc of America. Three-terminal offline monolithic switching integrated circuit TOP (three-terminal offline) uses PWM controller and power switch MOSFET. Combining two into one package has become the mainstream of the development of switching power IC. Therefore, according to the circuit specifications, it is suitable to use TOP switching integrated circuit to design switching power supply and use TOP100Y as the core to produce a voltage output of +12 V and +5 V with a power of about 4 W. Figure 15.2 is the internal circuit schematic diagram of TOP100Y [2]. Its main working principle is that the bias voltage is supplied by the control voltage VC to the parallel regulator and gate driver, the control terminal CONTROL

DRAIN

VC 0

Turn off/auto Restart circuit

+ 5.7V

+

5.7V 4.7V

1

Internal power supply

÷8

-

Parallel regulator/ Error amplifier

Power on reset

+

External trigger

QR QS

System shutdown

IC

VI LIMIT

Hot shutdown in sheet

Gate drive

oscillator DMAX CLOCK

-

SAW

+

Q QR

S

PWM Modulator

RE

CA

Internal circuit schematic diagram

Fig. 15.2 Internal circuit schematic diagram

Front locking

Minimum pass state time delay

SOURCE

150

DMAX

Self starting 自启

IB Slope=PWM Modulator gain-16%（mA）

Duty c ycle（%）

Fig. 15.3 Duty cycle and c-terminal current diagram

C. Yong et al.

DMIN ICD1 2.5

6.5

45

IC (mA)

current Ic is changed, the capacitor CA voltage is changed and the pulse width is adjusted by comparing with the PWM modulator. The main control circuit and gate driver control MOSFET, adjust the duty cycle of pulse, when the feedback current Ic↑ → D↓ → UO↓, the output UO will remain unchanged, and the secondary output voltage of the transformer will be stable. It should be noted that the C-terminal voltage does not exceed 9 V and the current does not exceed 100 mA. For the chip, the pulse width modulation is mainly accomplished by the pulse width modulator, and the gain of the pulse width modulation reflects the system performance, as shown in Fig. 15.3. TOP Switch-I has 8 dividers (÷8), so that every self-starting, S is sucked from zero to one end, the control end is supplied by an internal power supply, the control side bypass capacitor C is turned on after eight charging and discharging cycles, and the duty cycle is reduced to 5%, which reduces chip power consumption. TOP100Y has an over-current protection circuit. It is generally believed that when the current is too high, the chip will start itself. But when the chip is started, the current spike is too large, and the chip has the function of too large pulse front spike, which makes the circuit automatically restart without misoperation.

15.3 System Structure and Working Mode Usually, in the design of dual-output single-chip switching power supply, three-stage high-frequency transformer is needed [3]. In order to simplify the driving circuit, the original circuit components, the volume of power supply, and costs are reduced. In this design, an improved single-ended flyback structure is adopted, with a secondary coil N2 output +12 V and a feedback coil N3 output +5 V. The circuit structure is shown in Fig. 15.4. When the switching power transistor inside TOP100Y is turned on, the voltage on N1 is upper positive and lower negative, the polarity of induced voltage on N2 is upper negative, and lower positive, VD2 is cutoff, and N1 stores energy. When the switch power transistor in TOP100Y is cutoff, the induced electromotive force

15 The Design of Two-Channel Outputs Switching Mode … VD2 UF5402

151 L1 3.3uH Uo1 +12V

Ui 24V

N2

C1 2200P

R1 4.7K

N1 VD1 UF4005

VDZ 10.8V

C2 330uF

RL C3 100uF

N3 VD3 UF5402 R2 10 C4 100uF 3.3uH

L2

C5 100uF

IC1

IC2 P521

Uo2 +5V

TOP100Y C6 47uF

Fig. 15.4 Main circuit structure

on N1 is upper negative and lower positive, the polarity of induced voltage on N2 is upper positive and lower negative, and VD2 is on. After C2 discharge, L1 and C3 filters, the output voltage of +12 V is provided. R2 is the current limiting resistance of LED in IC2. The working current of LED is very small, so the voltage drop on R2 can be neglected. So there are: Uo1 = Uz + UF = 10.8 V + 1.2 V = 12 V Among them, UF is the positive voltage drop of the LED in the photocoupler, which is about 1.2 V. In order to improve the efficiency of high-frequency rectification and reduce the loss, VD2 chooses the Schottky diode or ultra-fast recovery diode. Induced AC voltage on N3 is filtered by the VD3 rectifier, C4 discharge, L2 and C5 to obtain +5 V output. At the same time, it supplies power to the collector of phototransistor in the photoelectric coupler. The photoelectric transistor emitted by the LED in the photocoupler P521 sends the emitter current to the controller of TOP100Y to adjust the duty cycle.

152

C. Yong et al.

Principle of voltage stabilization: Uo1↓ → Uo10;i--); }while((IFG1&OFIFG)!=0); }

Serial port communication is adopted between GPRS communication module and information perception module to initiate serial port setting (Fig. 20.3).

196

W. Han et al.

System software platform implementation

Fig. 20.3 Software platform topology

voidInitUSART(void) { P3SEL |= 0x30; //0011 0000 P3.4,5 = USART0TXD/RXD P3DIR |= 0x10; //0001 0000 ME1 |= UTXE0 + URXE0; // Enable USART0 TXD/RXD UCTL0 |= CHAR; // 8-bit character UTCTL0 |= SSEL0; // UCLK = ACLK UBR00 = 0x06; // 32k/4800 - 3.41 UBR10 = 0x00; UMCTL0 = 0x6F; // Modulation UCTL0 &= ~SWRST; // Initialize USART state machine IE1 |= URXIE0; // Enable USART0 RX interrupt System software platform implementation

20.4 System Highlights—Remote Control Through data collection and analysis, users can realize remote control of greenhouse equipment according to crop requirements. The user interface is shown in Fig. 20.4. Control equipment mainly includes fan, fill light lamp, drip irrigation, shutter, and sunshade. The user can judge whether water should be carried out according to the soil moisture condition and whether the fan should be switched on or off according to air temperature and humidity. According to the numerical value of the light intensity,

20 Intelligent Greenhouse Information Collection …

197

Fig. 20.4 The remote terminal controls the mobile terminal interface

determine whether the operation of shading or filling the light should be carried out. These users can be done with one click of a smart mobile terminal. First, the mobile terminal shall actively request to establish socket connection to the server with the specified IP address and port, and then transmit the instruction to the upper computer, and the instruction received by the upper computer is shown in Fig. 20.5. The control instructions transmitted are agreed in advance, such as AABB for fan opening control and BBAA for closing control. For supplementary light, CCDD is on and DDCC is off. When the upper computer receives the relevant instruction, the instruction will be processed according to the instruction content. The upper computer then transmits the processed instructions to the CPU of the intelligent sensing terminal through the GPRS wireless communication network, and finally realizes the control of the remote device.

198

W. Han et al.

Fig. 20.5 The process of converting user instructions to electrical signals

20.5 Conclusion The system is proved to be stable, reliable, real time, convenient to use, intelligent, and universal. The intelligent collection and control system of greenhouse based on the Internet of Things can realize the collection of air quality factors such as temperature, humidity, carbon dioxide, and environmental factors such as the amount of light and create the best environment suitable for the growth of crops through the corresponding control of the microcontroller. Remote communication realizes the communication between the field acquisition terminal and the remote control terminal of greenhouse greenhouses, which makes the management of greenhouse greenhouses more modernized. The system has low power consumption, simple operation, high expansion ability, and high efficiency, so it has a broad market application prospect. Acknowledgments The project is funded by Key R & D projects of Shandong Province (2019GNC106103).

References 1. Han, Y., Zhang, X., Liu, P., Chen, J.: Research on intelligent measurement and control system of Internet of Things in greenhouse. Agric. Technol. Equip. (2), 18–20, 24 (2017) 2. Zhang, Y., Yu, Q., Liu, P., Jiang, X.: Construction of intelligent control system for solar greenhouse. Anhui Agric. Sci. 46(15), 175–179 (2018) 3. Zhang, Y., Liu, P., Ma, H., Yu, Q., Dong Chen, F.: A distributed system architecture based on JSON. Chin. J. Agric. Mech. 36(5), 255–257, 266 (2015) 4. Shi, L., Chen, Z., Zhihua Gai, F.: The application of Internet of Things in intelligent agriculture. Agric. Mech. Res. 250–252 (2013) 5. Daoliang Li, F.: The Internet of Things and smart agriculture. Agric. Eng. 1–7 (2012) 6. Wang, H., Qin, G., Lv, Z., Zhu, S., Cai, Y.: Design of intelligent monitoring network system for agricultural greenhouse based on single-chip microcomputer. Wirel. Interconnecti. Technol. 17(5), 49–51 (2020)

20 Intelligent Greenhouse Information Collection …

199

7. Liu, Y., Gao, C.: Regulation regulation of tomato cultivation in intelligent greenhouse in north China based on Internet of Things. North Gard 2, 138–142 (2020) 8. Gao, H., Zhu, S., Chang, G., Lingfeng, F., Huang, Z.: Architecture and implementation of intelligent greenhouse system based on internet of things in agriculture. Res. Agric. Mech. 40(01), 183–188 (2018) 9. Hongchuan Chen, F.: Research on the development path of modern agriculture under the background of “Internet +”. Guangdong Agric Sci 42(16), 143–147 (2015) 10. Li, Y.: Intelligent measurement and control system of greenhouse tomato growing environment based on Internet of Things. Shandong Agricultural University (2019)

Chapter 21

Design and Key Technology Research of Civil-Military Integration Regulations and Standards Management System Tianming Huang

Abstract In this paper, the design scheme of the management information system of civil-military integration regulations and standards is described, and the key technologies in the system design are discussed. The system not only optimizes the workflow of the management of civil-military integration laws and regulations, but also realizes the information management of the whole process of the compilation and revision of civil-military integration laws and regulations. The use of the system improves the standardization and timeliness of civil-military integration regulations and standards management, improves the accuracy and traceability of regulations and standards data, and users respond well.

21.1 Introduction In recent years, the state and the military have issued some important laws, administrative regulations, and policy documents, which have a positive impact on promoting private enterprises to participate in the military information construction, accelerating the mutual transformation of military and civil information technology achievements, and enhancing the development vitality of the information industry. However, some of the existing laws and regulations are independent, and the policy system is selfcontained, without mutual support and complementarity. First, the system is not strong. National laws, military laws and regulations, and relevant policies are not closely connected, and the upper and lower supporting facilities are not enough [1]. The establishment of a complete legal and regulatory system is the basis for effective supervision of the development of civil-military integration. In order to continuously improve the regulatory basis, we must start to establish the investigation, research, preparation and feedback of laws, regulations, guidelines, standards and related regulatory and technical documents in the field of civil-military integration, so as to provide technical support and guarantee for the national civil-military T. Huang (B) College of Arts and Law, Wuhan Donghu University, Wuhan, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_21

201

202

T. Huang

integration policy and regulation. With the gradual establishment of the civil-military integration legal system, the original manual management mode can not meet the business needs. Therefore, it is necessary to carry out the civil-military integration legal standard management system project construction.

21.2 Overall Design 21.2.1 Functional Requirement The management system of civil-military integration laws and regulations needs to realize the following functions: the whole process management of the preparation, revision and release process of laws and regulations in the form of projects, the management of review meetings in the preparation and revision process of laws and regulations, the management and utilization of the issued laws and regulations, the management of the preparation and revision plan of laws and regulations, and the preparation and revision of laws and standards. The personnel involved in the process shall be managed, and the units involved in the preparation and revision of laws and standards shall be managed. Different functions can be provided to users with different permissions through permission configuration [2].

21.2.2 System Framework The system is based on J2EE architecture, mainly for the management of the preparation and revision process of laws and standards and the system of laws and standards. The overall framework is divided into three layers, as shown in Fig. 21.1. The presentation layer adopts Ajax technology, and the front page is based on Ext JS and div framework [3]. Through the flexible application of Ajax technology, the interaction and verification of business data between the front and back stations are realized, and the asynchronous loading and display function of data are realized. Through the use of Ext JS framework, the data set display function is improved in an all-round way, and the functions of data column position adjustment, column width adjustment, sorting, freezing, unfreezing, etc. are supported. The associated detailed data can be displayed through embedded tables. By calling the div style presentation form, the multi-level switch of page data is realized, which ensures the efficiency of the presentation page and optimizes the function and performance of the page. According to WFMC standard, the application layer encapsulates the workflow engine in components, which conforms to BPMN 2.0 standard. It supports the development of graphical business process and the definition of visual form. Users can easily call the interface to develop their own process processing related pages.

21 Design and Key Technology Research of Civil-Military … Fig. 21.1 Overall framework of the system

Ajax

203

Extjs

DIV

Presentation layer

project management

Meetings management

Plan management

Personnel management

System management

…...

Workflow engine

Application layer

JDBC

Database Data layer

The data layer is implemented based on JDBC. It establishes connection with database by calling JDBC API, sends the statement of operating database, and processes the result.

21.2.3 Presentation Layer Design The system page style is simple, using light blue as the main color, using the left and right split layout. The left page is a scalable menu at all levels, and the menu area can be hidden as required; the right page is the main work area, and the main work area supports multiple tabs switching, which is convenient for user operation. The system provides the front page of the system that can release information, including management requirements file download, the latest announcement, etc. The data editing page provides various input controls to facilitate the user to enter information, and provides data verification function to verify the validity of the entered data and return prompt information. The process processing page can graphically display the records of the whole process of preparation and revision of regulations and standards, including the current node status, processing time, processing personnel, processing opinions, submission documents, and other contents of each node processed, as well as the relevant information of the node to be processed. The data query page provides cross-module interactive query functions. For example, the corresponding project process files can be queried by project name and number, stage and file name association, and the published regulatory standard

204

T. Huang

files can also be queried. Level by level query can be realized, for example, from focusing on a project to focusing on a stage in the project, and then focusing on a step in a stage.

21.2.4 Functional Design The system consists of seven modules, including project management, conference management, system management, and preparation plan management. The specific content is shown in Fig. 21.2, and the support and call relationship among the modules is shown in Fig. 21.3.

Regulations and standards management

Meeting management

Project management

System management

Basic project information management

Project process management

Project version management

Preparation plan management

Expert management

Personnel management

Unit management

System management

Staffing management

Fig. 21.2 System function module design

Management system of laws and standards for civil military integration Project management

Meeting management

Preparation process management

Conference configuration management Collaboration unit management

Preparation plan management Project basic information management

Inclusion of external documents

Personnel management

System management Version management

Fig. 21.3 Function module relationships

Conference results management

21 Design and Key Technology Research of Civil-Military …

205

21.2.5 Data Layer Design 21.2.5.1

Logic Design

According to the division of business functions, the data layer of the system includes four categories: basic data, user authority data, process-related data, and business data, with a total of 33 data tables. The business data and process data store the core data of the platform. The business data can be further divided into six categories of subject data: project information, meeting information, system information, plan information, personnel information, and collaboration unit information. The process data records the process data of the preparation and revision process of laws and regulations. The database design mainly follows the third paradigm design idea cabinet. In the case of avoiding repetitive redundancy, a small number of derivative redundant fields are added. On the basis of the third paradigm, the standard is appropriately lowered to improve the response speed [4].

21.2.5.2

Physical Design

The database uses the object-relational database Oracle to store and manage data. The memory of Oracle can be divided into the system global area and the process global area, that is, system global area (SGA) and private global area (PGA), according to the sharing and private perspective. Reasonable memory management can greatly improve the performance of the data layer.

21.3 Key Technology Research 21.3.1 Support Process Consolidation and Splitting In the process of project implementation, if it is found that the current approval process of the project does not meet the actual requirements, the project can be marked as “special split” or “special consolidation” project. Select an item in the list to be split or merged, and execute “process split” or “process merge”. The split or merged item will start a new process. The new process automatically copies the process data of the original process, including the processing data and file data of each node. The original process is terminated. The new process is associated with the process before splitting or merging through the internal serial number. In the process of splitting or merging, you can append the memo information to the memo item of the split or merged project. After splitting or merging, enter the basic information of the project after splitting or merging manually in the to-do list to query the project and its associated information before splitting or merging.

206

T. Huang

21.3.2 Support Document Version Control A regulatory standard document needs to go through different stages in the project process from the first draft. In each stage, several document versions can be generated. The system needs to manage and control all document versions. After the project enters a new stage, the generated regulatory and standard documents are the current latest version, and the old version is retained. The current latest version is used by default for the files involved in processing in the process. Click the “version details” link next to the latest version to browse the historical version in the version details browsing interface, view and download the historical version files, and the owner of the file can also delete the invalid historical version files. The documents involved in version control include the main documents of all stages, preparation instructions of each stage, modification instructions of the three meetings, and other relevant documents. The definition rule of version number is phase number + modification number + 8-digit date.

21.3.3 Support Process Page Customization 21.3.3.1

Participants of Each Process Node can be Configured

Participants of each node in all stages of the system process can be specified through configuration operations. First, when the system is authorized, the system administrator can configure the system roles associated with each node, and add people to the roles. After a project is started, the secretary group administrator of the project can select one or more personnel from the associated roles of each node to participate in the specific operation of the node, and select from the three approval modes of “parallel approval”, “serial approval”, and “crazy one approval”. Through the above two authorization steps, the tight coupling relationship between personnel and process nodes is eliminated, the process modification is convenient, and the flexibility of the system is increased.

21.3.3.2

The Files Involved in Each Process Node can be Configured

All stages in the process and the files involved in all nodes in each stage can be freely configured by users. Configuration information includes document quantity, document identification number, document type, document reference name, etc. where document type includes approval, upload, browse, and meeting. After configuration, the meeting type file can only be used by the meeting management module. The approval type file displays the corresponding approval button and opinion input box on the processing page of this node, the upload type file displays the upload button and file selection box, and the browse type file displays the file

21 Design and Key Technology Research of Civil-Military …

207

download link. In order to simplify user operation, the file configuration information of a node that has been configured in the same phase can be directly used by other nodes in this phase.

21.4 Conclusion The management system of civil-military integration regulations and standards is a necessary tool and means for the management of the preparation and revision process of regulations and standards as well as the management of regulations and standards system. It provides a convenient control platform for business personnel, simplifies user operation, standardizes the business process of the preparation and revision of regulations and standards, and runs through the whole life cycle management of civil-military integration regulations and standards. Acknowledgments This work was supported by the grants from Youth Foundation WuHan Donghu University (2018dhsk007).

References 1. Ping, C.: Reflections on strengthening the construction of laws, regulations and mechanisms for the integrated development of military and civilian military information systems. Natl Def 8, 21–24 (2014) 2. Yi, S., Zhengyu, C., Min, S., Yuyan, X.: Design and implementation of enterprise legal risk prevention and control system information system. Sci. Technol. Bull. 34(09), 263–267 (2018) 3. Tianming, Y.: Research on enterprise legal affairs information system based on SOA architecture. Inform. Syst. Eng. 08, 54–56 (2017) 4. Jianhui, X., Lianmeng, J., Chao, X.: Design and implementation of policy and regulation information platform based on Web. Microcomput. Appl. 30(2), 4–6 (2018)

Chapter 22

A Reliable Wireless Monitor and Control System with Low Power for Greenhouse Microclimate Zhenfeng Xu, Jingjing Yin, and Xiujuan Li

Abstract The quality and energy consumption of wireless communication are the key factors that restrict the wide application of wireless sensor networks in greenhouse system. A reliable wireless monitoring and control system with low power for greenhouse was designed. The acknowledgement (ACK) mechanism was adopted to improve the wireless communication quality. The low power listening (LPL) technology was used to reduce the energy consumption of wireless nodes. In addition, the software watchdog was adopted to improve the anti-jamming capability of nodes. Test results show that there is no data packet loss during wireless transmission process. The current of nodes by using the LPL technology is much lower than that under the three working states of transmit, receive, and idle. The wireless monitoring and control system can meet the actual requirements of greenhouse microclimate.

22.1 Introduction The real-time monitoring and control of greenhouse microclimate is very important for ensuring the normal growth of crops. In the traditional monitoring and control system, sensors are all wired, which leads to many disadvantages, such as complex wiring, high cost, both poor mobility and poor scalability of sensor nodes. The application of wireless sensor network to the greenhouse environment monitoring system can overcome many disadvantages of that wired systems [1]. Many scientists and technicians have carried out the related research [2–4]. However, the defects of Z. Xu (B) · X. Li Hefei University, Hefei, China e-mail: [email protected] X. Li e-mail: [email protected] J. Yin Anhui Vocational College of Defence Technology, Lu’an, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_22

209

210

Z. Xu et al.

wireless sensor networks are also emerging, for example, the poor wireless communication quality and the short working life caused by the high-energy consumption of sensor nodes. These factors restrict the application of wireless sensor networks in the monitoring and control of greenhouse microclimate. Many researchers studied the wireless communication quality. Zhu designed a wireless monitoring system based on CC2530, and used it under three kinds of environment: a greenhouse, an open farmland, and an orchard [5]. The test results show that the average packet loss rate of wireless transmission is 7.6%. Li designed a wireless monitoring system with 10 sensor nodes based on MSP430 microprocessor and CC2420 wireless transceiver for greenhouse environment [6]. The test results show that the correct transmission rates of seven sensor nodes are more than 90%, and that of the rest two nodes are less than 70%. Hu designed a greenhouse wireless sensor node based on CC2430, and the highest correct data transmission rate was 96% [7]. For the environmental factors, the occasional data loss has little effect on system performance, because the missing data can be estimated by interpolation and prediction [8, 9], if necessary. However, for the control commands and the status information of equipment, reliable wireless communication is necessary. If such data is lost, the equipment cannot be switched on or off in time, which may cause indoor environmental factors to exceed the set ranges and excessive energy consumption. Therefore, the control command and the status information of equipment have a high requirement for the wireless communication quality. Besides the wireless communication quality, the energy consumption is another key factor that restricts the wide application of wireless sensor networks. Most of the energy consumption of nodes is used for wireless communication. The power consumption of nodes when they are in the dormant state is obviously less than those when nodes are in the transmitting state, the receiving state and the idle state [10], so the wireless module should be closed in time to reduce energy consumption in order to prolong its working life when no wireless communication is needed. In view of the above problems, we designed a reliable greenhouse wireless monitoring system with low power by adopting the acknowledgement mechanism and the low power listening technology.

22.2 Realization Principles Although some protocols, such as TDMA, CSMA, etc., are adopted in the MAC layer of wireless sensor networks to reduce the packet loss caused by collision and other factors [11], the wireless communication quality is still worse than that of wired communication, especially when wireless communication is frequent. The acknowledgement (ACK) mechanism is an effective way to improve the wireless communication quality [12]. The basic process of ACK mechanism is described as follows. At first, the source node sends a packet to the objective node. When the objective node receives the packet, it sends an ACK message to the source node. If the source node receives the ACK message, it indicates that the packet is sent

22 A Reliable Wireless Monitor and Control System … Source node Objective node

Preamble Listening

211 Data

Listening

Data

Listening

Fig. 22.1 The working process of LPL technology

successfully, and the source node does not have to send the packet again. Otherwise, if the objective node does not receive the packet, accordingly, the source node does not receive the ACK message. Then the source node will resend the packet until the objective node receives or the number of transmission times has reached the set limit. The maximum number of transmission times must be set for the ACK mechanism, which can prevent the source from sending packets endlessly without receiving the ACK message. Assuming the packet loss rate of single wireless transmission is 50%, and the maximum number of transmission times is 10, then the packet loss rate can be reduced to less than 0.1%. It can be seen that the ACK mechanism can greatly improve the communication quality. Because the power consumption of nodes when they are in the dormant state is obviously less than those when nodes are in other states, the nodes should be in the dormant state as far as possible when there is no data to send, which can pro-long the working life of the node. The low power listening (LPL) is an effective way to solve this problem [13]. The working process of the LPL between a wire-less source node and an objective node is shown in Fig. 22.1. The objective node periodically sends listening signals to detect radio frequency (RF) activity. If there is no RF activity, then the node returns to the dormant state. If RF activity is detected, the objective node remains active and begins to receive data packets. The source node first sends a preamble when it is necessary to send data. Because the objective node is in the listening/sleep alternation states, it is required that the length of the preamble sent by the source node is longer than a listening/sleep period of the objective node. After listening to the preamble, the objective node sends an ACK message to the source node. After receiving the ACK message, the source node begins to send data packets. When there is no data transmission, both the source node and the objective node are in the dormant state. For the objective node, each listening period is very short, and the duty ratio of wireless communication on/off can be reduced to less than 1%. Therefore, the objective node is in dormant state most of the time. For the source node, the number of packets that need to be sent is usually not much, so the source node is also in dormant state most of the time. Therefore, the LPL technology can reduce the energy consumption of nodes significantly.

212

Z. Xu et al.

22.3 Wireless Monitoring and Control System for Greenhouse Microclimate A wireless monitoring and control system for greenhouse microclimate was designed as Fig. 22.2. The system includes two indoor sensor nodes, one outdoor sensor node, one control node, one relay node, one base station, and one industrial computer. The functions of each node are described as follows. The indoor sensor nodes are used to collect indoor air temperature, relative humidity, and illuminance. The outdoor sensor node is used to collect outdoor air temperature, relative humidity, and illuminance. The data of both indoor and outdoor environmental factors acquired by sensor nodes are transmitted to the base station through the relay node, and then enter the industrial computer through RS-232 serial port. Data are processed, stored, displayed, and uploaded to the Internet by the industrial computer. The base station receives control commands from the industrial computer, and sends them to the control node. After receiving the control commands, the control node realizes the on-off control of the indoor equipment, and returns the state of the equipment to the base station and the industrial control computer. The Mica2 nodes are taken as the core of all wireless communication nodes, and then different sensors and other components are added to accomplish different functions. The low power microprocessor Atmegal28L and the wireless transceiver CC1000 are adopted in the Mica2 node. The node works in the 433 MHz band, and has stronger diffraction ability compared with the 2.4 GHz communication band. For the sensor nodes, the SHT11 temperature and humidity sensors and TSL2561 illuminance sensors are adopted. Because the measurement range of TSL2561 cannot meet the requirement of illumination, a glass plate with a transmittance of 20% is added to the top of the sensor. For the control node, the AC contactors driven by the triode are adopted, in order to realize the on/off control of the equipment. There are four AC contactors installed on the control node, which are used to control four kinds of equipments. For the relay nodes, no need to add any device and its function is to complete the relay transmission of environmental data and control commands. For the base station, the RS-232 interface circuit is designed, in order to realize the communication between the wireless monitoring network and the upper computer. The circuits of sensor nodes, relay node, and control node are shown in Fig. 22.3.

Outdoor sensor node Indoor sensor nodes Relay node

Base station

Control node Fig. 22.2 The architecture of the wireless monitoring and control system

Internet

22 A Reliable Wireless Monitor and Control System …

(a) Sensor node

(b) Relay node

213

(c) Control node

Fig. 22.3 The circuits of various wireless nodes

Node software is developed based on the TinyOS operating system, which is an open-source operating system designed specifically for embedded wireless sensor networks. The main task of sensor nodes is to achieve periodic acquisition of environmental factors. When the node is powered, the wireless communication module, TSL2561 and SHT11 sensors are launched in turn, and a cycle timer is set up. Then, after each timing, the sensor nodes collect the data of environmental factors and watchdog is set up. The watchdog is started when the microprocessor begins to collect the data of environmental factors. Under normal circumstances, the microprocessor can complete data collection in feeding dog time. After reading, the watchdog is turned off. If the microprocessor cannot collect data in time because of interference and other factors, then the system will reset automatically when the watchdog timer ends, which ensures that the nodes have strong anti-interference ability. The main task of the control node is to receive the command message from the base station, realize the on/off control of the indoor equipment, and then send the state information of the equipment to the base station. After receiving a command message, the command node mainly completes two tasks based on its analysis. One is to identify which equipment the command is aiming at. The other is to achieve the control (on or off) of the objective equipment. In order to improve the antiinterference ability and the wireless communication quality, the software watchdog and ACK mechanism are also used in the data transmission process. Compared with the sensor nodes and the control node, the task of the relay node is much simpler, that is, to receive and transmit all kinds of data. The software development of the relay node focuses on the realization of the LPL technology. Of course, the ACK mechanism is still used in data transmission. The tasks of the base station node are as follows: (1) receive all kinds of data transmitted by the relay node and transmit them to the PC monitoring terminal; (2) receive the control commands sent by the PC monitoring terminal, and send them to the control node through the relay node, thus realizing the switch control of the corresponding equipment. Both the ACK mechanism and the LPL technology are also adopted.

214

Z. Xu et al.

22.4 Experiments 22.4.1 Experimental Scheme Firstly, the wireless communication quality of the control node is tested, because the transmission of control command and equipment state information requires the highest communication quality. A serial debugging tool on the PC is used to send control commands, and the control commands are sent to the base station by the RS-232 serial port. After receiving the control commands, the base station transmits them to the control node. After receiving the control commands, the control node identifies the equipment number and the control commands, and switches the corresponding AC contactor on or off. Then, the state information of the AC contactor is obtained by the control node, and then is sent to the base station. Finally, the state information is displayed on the serial debugging tool interface. Each equipment is repeatedly turned on and off 50 times, and there are 200 times operation. The number of equipment state information displayed on the serial debugger can reflect the wireless communication quality. In order to test the impact of the ACK mechanism on the wireless communication quality, two testing schemes are set up. The first scheme does not use the ACK mechanism, and the second scheme uses the ACK mechanism. The control node in the two schemes is tested, respectively. The low power consumption is mainly tested on the relay node, because the node is powered by a solar panel. The transmit power of the relay node is 10 dBm. The relay node is set in the following four states in turn: transmit state, receive state, idle state, and LPL state. In accordance with convention, the power consumption of the node is checked by measuring its working current, because the working voltage is constant. In the test, the wake-up period in the LPL is set to 0.5 s.

22.4.2 Results and Analysis In the two test schemes, the number of command execution times (the control commands are successfully received by the controller node) and that of the equipment states (the equipment state information received by the monitoring terminal) are shown in Table 22.1. The results show that the control command and equipment state information are lost during the wireless transmission process in scheme 1, which is obviously difficult to meet the requirements of the two kinds of data for the wireless communication Table 22.1 The number of successful data transmission in the two test schemes

Schemes

Control command

Execution

Equipment state

1

200

192

187

2

200

200

200

22 A Reliable Wireless Monitor and Control System …

215

Table 22.2 The working current of the relay node in four working states Working states

Transmit

Receive

Idle

LPL

Current (mA)

32

31

28

3

quality. In scheme 2, no data packet is lost. The wireless communication quality is improved significantly by using the ACK mechanism. The working current of the relay node in four working states is shown in Table 22.2. The measurement results show that the current in the LPL state is much less than that in other working states, which is very important for the wireless nodes powered by battery or solar energy.

22.5 Conclusions A reliable greenhouse wireless monitoring and control system with low power has been developed. The use of ACK mechanism significantly improves the wireless communication quality, and can ensure reliable and stable transmission of all kinds of data. The use of LPL technology reduces the power consumption of wireless nodes, which is of great significance for battery or solar-powered nodes. The system can meet the monitoring requirements of greenhouse microclimate and has a good practical value. Acknowledgments This work was supported by the Support Program of Outstanding Young People in Universities of Anhui Provence (gxyqZD2018134) and the Scientific Research Fund of Talent Introduction in Hefei University (18-19RC35).

References 1. Mirabella, O., Brischetto, M.: A hybrid wired/wireless networking infrastructure for greenhouse management. IEEE Trans. Instrum. Meas. 60, 398–407 (2011) 2. Wang, N., Zhang, N., Wang, M.: Wireless sensors in agriculture and food industry–recent development and future perspective. Comput. Electron. Agric. 50, 1–14 (2006) 3. Ahamed, V.S.J.S.F.: Smart wireless sensor network for automated greenhouse. IETE J. Res. 61, 180–185 (2015) 4. Kumar, S.A., Ilango, P.: The impact of wireless sensor network in the field of precision agriculture: a review. Wirel. Pers. Commun. 98, 685–698 (2018) 5. Zhu, B., Han, W., Wang, Y., Wang, N., Chen, Y., Guo, C.: Development and evaluation of a wireless sensor network monitoring system in various agricultural environments. J. Microw. Power 48, 170–183 (2013) 6. Li, Z., Ning, W., Hong, T., Wen, T., Liu, Z.: Design of wireless sensor network system based on in-field soil water content monitoring. Trans. Chin. Soc. Agric. Eng. 26, 212–217 (2010) 7. Hu, P., Jiang, T., Zhao, Y.: Monitoring system of soil water content based on ZigBee wireless sensor network. Trans. Chin. Soc. Agric. Eng. 27, 230–234 (2011)

216

Z. Xu et al.

8. Xie, Y., Chen, X., Zhao, J.: A double weighted LS-SVM model for data estimation in wireless sensor networks. J. Beijing Instit. Technol. 21, 134–139 (2012) 9. Yan, N., Zhou, M.Z., Tong, L.: An estimation algorithm for missing data in wireless sensor networks. Int. J. Smart Sens. Intell. Syst. 6, 1032–1053 (2013) 10. Sun, L., Li, J., Chen, Y., Zhu, H.: Wireless Sensor Network. Tsinghua University Press, Beijing (2005) 11. Liu, Q., Chang, Y., Jia, X.: A hybrid method of CSMA/CA and TDMA for real-time data aggregation in wireless sensor networks. Comput. Commun. 36, 269–278 (2013) 12. Hu, S.Q., Wang, J.F.: Smart low power listening for wireless sensor networks. J. Commun. 30, 95–101 (2009) 13. Jurdak, R., Baldi, P., Lopes, C.V.: Adaptive low power listening for wireless sensor networks. IEEE Trans. Mobile Comput. 6, 988–1004 (2007)

Chapter 23

Prediction of Excessive Cadmium in Rice Based on Weighted Bayesian Fusion Model Baohua Zhang, Wei Wang, Yi An, Yuan Jiao, and Yue Li

Abstract Soil Cd pollution directly affects the safety of agricultural products. Frequent Cd pollution incidents have repeatedly reminded that soil Cd pollution in China is not optimistic. In this paper, the linear regression prediction based on SOM, pH, and soil Cd in the data was firstly carried out. The fitting degree R2 was 0.109. The model was not accurate enough. Therefore, this paper converts the study of rice Cd beyond the predicted regression problem into a classification problem. According to the Chinese national standard GB2762-2012 food contaminant limit, the content of Cd in rice should not exceed 0.2 mg/kg. Finally, the adaptive classification algorithm based on the weighted Bayesian fusion model is compared with the classical classifier model (SVM, RF). The adaptive classifier is better than the classical classifier in F1 and accuracy index. The adaptive classifier model obtained from the above studies can not only guide the classification of rice Cd but also save huge costs in the monitoring and prevention of Cd pollution.

23.1 Introduction The rapid development of China’s industry and agriculture, Cd pollution has attracted the attention of the country, the typical representative of Cd-contaminated crops is rice. According to Guolian’s analysis, rice Cd pollution is mainly related to rice B. Zhang · W. Wang College of Computer Science, Nankai University, Tianjin 300350, China Y. An Agro-Environmental Protection Institute, Ministry of Agriculture and Rural Affairs, Tianjin 300191, China Y. Jiao Tianjin University of Commerce Boustead College, Department of Computer and Information Technology, Tianjin 300384, China Y. Li (B) Nankai University, Tianjin, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_23

217

218

B. Zhang et al.

planting pollution. Therefore, it is a very important question to study whether the Cd content of crops grown in soil is excessive. In this paper, the following studies on rice Cd exceeded the following criteria: (1) Based on the regression prediction of rice Cd content, (2) Classification based on whether the Cd content of rice exceeds the standard. A series of studies on rice Cd detection data were standardized, distribution characteristics, and pretreatment. The rice Cd content was predicted by a simple linear regression model, and the linear regression model R2 (goodness of fit) was 0.19. The value of R2 is small, indicating that the linear regression model is not ideally fitted. Therefore, according to the Chinese national standard GB2762-2012 food contaminant limit regulations, the Cd content in rice should not exceed 0.2 mg/kg, and the rice Cd prediction problem is transformed into the second classification problem of rice Cd exceeding the standard. The classical classifier models, such as Naive Bayes (NB) and Support Vector Machine (SVM) are selected for classification training. The accuracy and F1 are 62% and 72%, respectively. To further improve the prediction accuracy, this paper designs a weight-based design. The adaptive classification algorithm of the Bayesian fusion model has an accuracy rate of 65.8 and 76.1%. This is of great significance for the detection of excessive Cd in rice in agriculture. The main content and organizational structure of this paper are as follows: The first chapter introduces the background of the research. The second chapter mainly introduces the research status of rice Cd content prediction. The third chapter mainly introduces the weighted naive Bayesian model proposed in this paper. The fourth chapter mainly introduces the experimental results of various methods. The fifth chapter summarizes the research content and results of this paper.

23.2 Related Work The current models describing the migration and transformation of heavy metals in soil-crop systems can be divided into three broad categories. One type is expressed as an absorption coefficient or an enrichment factor of a direct object. He [1] and others calculated the contents of lead, Cd, zinc, and copper in soil and vegetables by enrichment coefficient in their articles. The absorption coefficient method is simple to operate, but the error is generally large and the evaluation accuracy is not sufficient. The second category is the mechanism model. Wu [2] divides the process of crops to absorb heavy metals into three parts, namely, the desorption of heavy metals from soil solids into soil solutions, the diffusion of heavy metals to the root surface through diffusion and mass flow, and the absorption of heavy metals by crop roots. Furthermore, a quantitative relationship is established to determine the amount of plant uptake based on soil solution concentration, buffer capacity coefficient, diffusion coefficient, total root length, average radius, plant transpiration, and growth time. Brennan [3] established a mechanism model for the absorption, migration, and accumulation of lead in soil-maize systems to absorb the mechanism model of absorption and migration of heavy metals from soil. The model sets a large number

23 Prediction of Excessive Cadmium in Rice Based …

219

of assumptions, and the parameters of the model are numerous, and the uncertainty is large. Sterckeman [4] uses the Barber-Cushman model to predict Cd uptake by corn, but the model requires more parameters and is computationally complex. The mechanism model requires more parameters, the measurement is more difficult, and the computational complexity makes it practically greatly discounted, which is difficult to promote and use [5]. The third category is the statistical model, the so-called test model. Empirical models such as the Freundlich model, stationary model [6], linear or linear regression models [7], have been widely used due to their fewer parameters, simple modeling, and high estimation accuracy. In the established soil-crop Cd transport model, the parameters mainly include soil Cd content, soil pH, soil organic matter or organic carbon, soil texture, and soil cation exchange capacity. With the deepening of research on the transport mechanism of Cd in the soil-crop system, the parameters of participation in model construction are also being improved, such as the model of effective state of Cd plant added to the soil, meteorological factors, and the accuracy of soil zinc increase model [8]. In recent years, some researchers have suggested that the historical accuracy (i.e., the historical probability of crop Cd exceeding the standard) can be further improved when assessing the migration of Cd in the soil-crop system [6].

23.3 Proposed Method 23.3.1 Bayes The naive Bayesian algorithm is a commonly used algorithm based on Bayes’ theorem. The application condition of the algorithm is that the features must be independent. Bayesian formula: P(A|B) =

P(B|A)P( A) P(B)

P(A) is the prior probability; P(B|A) is the conditional probability; P(A|B) is the posterior probability.

23.3.2 Naive Bayes Bayesian’s hypothesis is that there is only one feature in total, but in practical applications, very few things are only affected by one feature, and there are many factors that often affect one thing, so the final model function is

220

B. Zhang et al.

PN B (X ) = argmax P(C) c

n 

p(xi |C)

i=1

X = {x1 , x2 L , xn }, C = {c1 , c2 , Lck }

23.3.3 Weighted Naive Bayesian Fusion Model Since the independence of Naïve Bayes requirements are relatively strong, compared with most data sets, the independence hypothesis between attributes cannot be fully satisfied. Therefore, the idea of feature weighting is applied to Bayesian, and each conditional probability is given. The weight θ i is used to weaken the independence between attributes, and the formula for weighted Bayesian (WNB) [9] is P(C) PN B (X ) = argmax c

n 

p(xi |C)θi

i=1

X = {x1 , x2 L , xn }, C = {c1 , c2 , Lck } For the study of SOM interpolation data, the classification algorithms QDA, SVM, and Naïve Bayes, which have the highest baseline for the classification of rice Cd, have the classification accuracy of the data after removing the interpolation data SOM (pH + soil Cd). The increase of F1 value indicates that the interpolation data SOM has a negative impact on the classification effect. There are two solutions to this. One is to design an interpolation method that is more suitable for high overlap data such as rice Cd exceeding the standard; an adaptive algorithm can adaptively classify the interpolated data. The whole classifier is based on the weighted Bayesian [10] fusion classification model design. Naive Bayes uses the maximum likelihood method for parameter estimation. The whole algorithm is designed as follows: 1. First, set up two base classifiers A and B, which are used to learn non-interpolated data (represented by X) and interpolated data (represented by Y ). Now assume that A classifier learns non-interpolated data, B classifier learns interpolation data. 2. Assuming that the interpolated data and the non-interpolated data are considered to be independent of each other, then the logarithmic probability formula using naive Bayes has   P = log (PA (X ))θ1 · (PB (Y ))θ2

(1)

23 Prediction of Excessive Cadmium in Rice Based …

221

PA (X) represents the probability of non-interpolated data X on the A classifier, and PB (Y) represents the probability of the interpolated data Y on the B classifier, where θ 1 and θ 2 are two hyperparameters, according to previous studies, non-interpolation The classification effect of the data is much better than the classification of the interpolation data, so setting these two parameters to adjust, and then continuing to derive the formula (proposed θ 1) are   θ2  P = log (PA (X ))1 · (PB (Y ))θ1 θ 1

(2)

3. According to Eqs. 1 and 2, the probability PA (X) of the non-interpolated data X on the classifier A, the probability PB (Y) of the interpolated data Y on the classifier B, and the amount of data for the hyperparameter θ 2/θ 1 are obtained. When you are old, you can use the learning method, but when the amount of data is small, you can use the search method. The reason is that when the amount of data is small, the learned hyperparameters are basically over-fitting and untrustworthy, so when the amount of data is small, using the lookup method, the probability of classification is calculated according to Eq. 1 or Eq. 2. 4. Compare the probability sizes corresponding to each category to get the corresponding classification results. The above is the specific steps of the design flow of the adaptive algorithm. It can be seen that the overall process of the adaptive algorithm is a fusion model based on weighted Bayes [11], in which the two base classifiers A and B can be flexibly selected according to different data.

23.4 Experiment 23.4.1 Linear Regression Model The coefficient of determination R 2 : the coefficient of determination is used to measure the good or bad regression, and is the goodness of fit of the regression fitting curve. The larger the coefficient of determination, the better the goodness of fit. The calculation formula is 2 n  y−y SS R SS E = ni=1 =1− 2 SST SST − y) (y 1 i=1 

R2 = −

y : mean value. y: predicted value.

222

B. Zhang et al.

SST :

2

n



−

yi − y

i=1

SS R :

n



2



−

yi − y

i=1

SS E :

n 

2



yi − yi

i=1

SOM, PH, soil Cd content as an independent variable, rice Cd as a dependent variable, to fit the linear regression model, and finally, the value of the determination coefficient R 2 is 0.19, indicating that the linear regression model has a poor-fitting effect. The model cannot be used to predict whether rice Cd exceeds the standard.

23.4.2 Weighted Naive Bayesian Fusion Model SVM is used for the data of two characteristic attributes of non-interpolation data PH + soil Cd, and the adaptive classification algorithm of single feature SVM is used for the interpolation data SOM. The line graph corresponding to the accuracy and F1 value is shown in figure (a) and figure (b), respectively (where X-axis is the value of θ 2/θ 1, Y-axis is the accuracy and F1 value). As shown in Fig. 23.1, it can be seen that the accuracy fluctuates up and down with the changing trend of θ 2/θ 1, among which the value of θ 2/θ 1 with the best accuracy is between 0.2 and 0.4; the changing trend of F1 is also up and down, and the more close to 0, the greater the fluctuation is, and the several values of θ 2/θ 1 with the best F1 are between 0.2 and 0.4. Table 23.1 is the comparison of the best accuracy and F1 value of the adaptive classifier (SVM is used for PA (X ), and SVM is used for PB (Y )) with the baseline classifier (at this time, θ 2/θ 1 is 0.223). As shown in Table 23.2, for the adaptive classifier (SVM for PA (X ), SVM for PB (Y )) (accuracy is 0.6584, F1 value is 0.7612), the accuracy increased by 3.09% relative to the SVM baseline, increased by 3.54% relative to the QDA, and 4.42% relative to the Naïve Bayes. F1 increased by 3.68% relative to SVM baseline, 4.06% relative to QDA, and 5.35% relative to Naïve Bayes. The results show that the adaptive algorithm based on the weighted Bayesian fusion model is effective and superior.

23.5 Conclusion In this paper, by comparing the effects of linear regression model and common classifiers, an adaptive algorithm model based on weighted Naïve Bayesian fusion model is proposed and implemented to predict whether the content of rice Cd exceeds

23 Prediction of Excessive Cadmium in Rice Based …

223

Fig. 23.1 Line chart of accuracy and F1 value of SVM for non-interpolation probability and interpolation probability Table 23.1 The classification results of conventional classifiers Classification algorithm

QDA [12]

SVM [13]

GBDT [14]

KNN [15]

Random forest

Naive Bayes [11]

Accuracy

0.6230

0.6275

0.5110

0.5499

0.4923

0.6142

F1

0.7206

0.7244

0.5165

0.6309

0.4698

0.7074

224

B. Zhang et al.

Table 23.2 Comparison between adaptive classifier and baseline of QDA, SVM, and Naïve Bayes Operation

SVM

QDA

Naïve Bayes

Adaptive classifier

Accuracy

0.6275

0.6230

0.6142

0.6584

F1

0.7244

0.7206

0.7074

0.7612

the standard. To verify the validity of the adaptive classification algorithm, and the best combination of adaptive classifiers is the SVM of the interpolation data SOM the best. This result is very obvious for the case where the data is highly overlapped and inseparable, and the improvement effect is obvious. The effectiveness and superiority of the adaptive algorithm based on the weighted Bayesian fusion model are illustrated. Acknowledgments This work was supported by National Key Research and Development Program of China (2016YFB0201304), the Tianjin Natural Science Foundations (17JCYBJC23000), and National Key Research and Development Program of China under the grant number (2018hjyzkfkt002).

References 1. He, J., He, Q., Liu, D. et al.: Influencing factors and prediction models of hygienic safety threshold for soil cadmium foods: A case study of paddy soil in Changsha. Acta Pedol. Sin. 54(5) (2017) 2. Wu, Q.: A principle model for quantitative plant uptake of soil heavy metals. Acta Pedol. Sin. 1, 68–76 (1994) 3. Brennan, M.A., Shelley, M.L.: A model of the uptake, translocation, and accumulation of lead (Pb) by maize for the purpose of phytoextraction. Ecol. Eng. 12(3–4), 271–297 (1999) 4. Sterckeman, T., Perriguey, J., Caël, M., et al.: Applying a mechanistic model to cadmium uptake by Zea mays and Thlaspi caerulescens: consequences for the assessment of the soil quantity and capacity factors. Plant Soil 262(1–2), 289–302 (2004) 5. Tudoreanu, L., Phillips, C.J.C.: Modeling cadmium uptake and accumulation in plants. In: Donald, L.S. (ed.) Advances in Agronomy, vol. 84, pp. 121–157. Academic Press (2004) 6. Xu, J., Bo, W., Zhang, L., et al.: Risk assessment of excessive Cd of rice in Xiangtan, Hunan based on Bayesian method. J. Appl. Ecol. 27(10), 3221–3227 (2016) 7. Nan, Z., Li, J., Zhang, J., et al.: Cadmium and zinc interactions and their transfer in soil-crop system under actual field conditions. Sci. Total Environ. 285(1–3), 187–195 (2002) 8. Römkens, P.F.A.M., Guo, H.Y., Chu, C.L., et al.: Prediction of Cadmium uptake by brown rice and derivation of soil–plant transfer models to improve soil protection guidelines. Environ. Pollut. 157(8–9), 2435–2444 (2009) 9. Krauss, M., Wilcke, W., Kobza, J., et al.: Predicting heavy metal transfer from soil to plant: Pltential use of Freundlich-type funcitons. J. Plant Nutr. Soil Sci. 165, 3–8 (2002) 10. Huang, Y., Zhang, Y.: An improved weighted Bayesian malware identification method. J. Shenyang Ligong Univ. 38(01), 43–47 (2019) 11. Webb, G.I., Pazzani, M.J.: Adjusted probability Naive Bayesian induction. In: Australian Joint Conference on Artificial Intelligence. Springer, Heidelberg, pp. 285–295 (1998) 12. Siqueira, L.F.S., Araújo Júnior, R.F., de Araújo, A.A., Morais, C.L.M., Kássio, M.G.: Lima. LDA versus QDA for FT-MIR prostate cancer tissue classification. Chemometr. Intell. Lab. Syst. 162 (2017)

23 Prediction of Excessive Cadmium in Rice Based …

225

13. Bai, R., Wang, X., Wang, X.: Research on automatic indexing of sci-tech literature based on support vector machines and core feature words. Inf. Theory Pract. 37(07), 129–134 (2014) 14. Chen, X, Zhang, T., Zhang, R., Huang, F., Wang, Z., Zhao, Q.: Scene user identification method based on data mining algorithms such as GBDT. Telecommun. Sci. 34(S2), 167–175 (2018) 15. Sharma, S., Kumar, A., Gupta, V., Tomar, M.: Dielectric and ferroelectric studies of KNN thin film grown by pulsed laser deposition technique. Vacuum 160 (2019)

Chapter 24

Segmentation of Oilseed Rape Flowers Based on HSI Color Space and Local Region Clustering Jiahua Zeng, Xuan Wang, and Kaiqiong Sun

Abstract In order to improve the segmentation accuracy of rape flowers images under natural illumination, an automated segmentation method combing HSI spatial color threshold and local region clustering is proposed in this paper. The original image is converted into HSI color space. And then the candidates target region is located by color threshold. The image is converted back into RGB color image with background marked as black. At the same time, the original image is clustered within a local region using the distance defined in the LAB color space. The two results from the above process are combined into the last segmentation result, where the candidates target region is refined with the local clustering result. In order to verify the effectiveness of the proposed method, images of rape flowers obtained in the real environment are tested. The experimental results show that this method can accurately extract the target rapeseed from the image of the canola flower with complex background and strong illumination.

24.1 Introduction Oilseed rape is one of the most easily planted crops in China, and also one of the main sources of edible oil, high energy, and protein powder plants [1]. Flowering degree is a key stage in oilseed rape cultivation. It marks the transition from vegetative growth to reproductive development, which is sensitive to extreme temperatures, nutrients, and water. Therefore, flowering provides important information for formulating optimal crop management plans. To quickly quantify the flowering time or number of rape J. Zeng (B) · X. Wang · K. Sun Wuhan Polytechnic University, Wuhan, China e-mail: [email protected] X. Wang e-mail: [email protected] K. Sun e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_24

227

228

J. Zeng et al.

flowers, it required a method to detect the flowers of rape in the field under natural conditions. The objective of our work is to develop segmentation methods for flowers present in a scene captured with the normal digital camera of visible range. Under such imaging condition, Aggelopoulou et al. [2] proposed a color threshold method to detect flowers, which required images to be taken at a specific daylight time with a black screen behind the trees. Thorp et al. [3] proposed image analysis process that is based on threshold processing according to color (in HSI color space) and size features. Shuai et al. [4] used yellow template to remove background information and then used K-means clustering to segment rape flowers. A color similarity judgment model was established based on RGB color space, and the appropriate color similarity judgment threshold was determined to classify the canola flower image in [5]. HSI transform has been generally considered useful for color image segmentation under variable illumination conditions in outdoor agricultural scenes. Thorp et al. [3] statistically calculated the maximum and minimum hue, maximum and minimum saturation, maximum and minimum intensity of yellow flowers as boundary conditions for threshold segmentation to extract the target. It can separate the target from the background, but there is excessive segmentation. Partial backgrounds that are similar in color to rape flowers are located at the same time, so they need to be further segmented. For such problem, a method combining HSI spatial color threshold with local means clustering is proposed in this paper. The original image is converted into HSI color space. And then the candidates target region is located by color threshold. The image is converted back into RGB color image with background marked as black. At the same time, the original image is clustered within a local region using the distance defined in the LAB color space. The two results from the above process are combined into the last segmentation result, where the candidates target region is refined with the local clustering result.

24.2 Materials and Methods 24.2.1 HSI Color Space Threshold Segmentation Because that HSI color space transformation is beneficial to color image segmentation under the condition of variable illumination of outdoor agricultural scene, the outdoor rape image is transformed from RGB space to HSI space. After the image is transformed into HSI color model, the average maximum and minimum value of rape flower in H channel of HSI color space calculated in [4], u 1 , and u 2 are used to divide the image based on the following, S = (H > u 1 ) ∩ (H < u 2 )

(1)

24 Segmentation of Oilseed Rape Flowers Based …

229

Fig. 24.1 a Original image; b HSI threshold result; c SLIC super-pixel result; d Result of fusion of the two

where S denoted the segmented region, the H value is between the average maximum and minimum value. Figure 24.1b presents the result where the original color of pixels in Fig. 24.1a is kept within S, and the color of other pixels is set to black. However, the image segmentation results using HSI color threshold have the situation that the stamen is cut into the background, that is, there is a hole in the center of the flower, not a complete flower, which is far from the expected goal. Therefore, we need to further process the image.

24.2.2 SLIC Super-Pixel Segmentation Simple linear iterative clustering [6] is easy to implement image segmentation algorithm, which is usually used for image prepossessing. Firstly, the general color image is transformed into 5-D feature vector in LAB color space and position space coordinates, and then the standard of measuring distance is constructed to locally cluster image pixels. The implementation steps of SLIC are as follows:

230

J. Zeng et al.

(1) Initialize cluster center k. (2) Re-select the cluster center. In order to avoid the cluster center falling on the contour with large gradient, we need to re-select the cluster center in the n × n neighborhood of the cluster center (usually n = 3). (3) Set a class label for each pixel in the cluster center field, that is, which cluster center the pixel belongs to. (4) Measure distance. Distance consists of color distance and space distance. (5) Optimize by iteration. Repeat the above process until the clustering center of each pixel no longer changes. The result of Fig. 24.1a with the above super-pixel clustering is shown in Fig. 24.1c.

24.2.3 Segmentation Result Fusion In the binary image generated by HSI color threshold segmentation, we find that the hole problem and similar color background problem are difficult to solve. The super-pixel segmentation has a good adhesion to the contour of rape flower, so the two results are combined to solve the problem of HSI color threshold segmentation. In the super-pixel algorithm, there is a two-dimensional matrix describing the super-pixel labels of all pixels in the return value. It is denoted as S [x * y], and x * y as the space size of the image. After the HSI threshold segmentation is completed, a binary image with the same size as the original image can be obtained. It is denoted as Y [x * y], and the value 1 in the binary image represents the rape flower region. (1) Set three one-dimensional arrays, M [1… n], N [1… n], K [1… n], where n is the number of super-pixels in the super-pixel segmentation. (2) Count the number of pixels in each super-pixel in S, and store the data into the array M, that is, M(i) is the number of pixels in super-pixel i. Every element in M is greater than 0. (3) Count the number of target pixel number of each super-pixel in the Y image covered by S, that is, count the total number of pixels of the target rape flower in each super-pixel, and store the value in the array N, so N[i] represents the number of pixels of the target in super-pixel i. The element in n might be 0. (4) After getting M[i] and N[i], keep the ratio of N(i) and M(i) in array K, that is, K (i) = N (i)/M(i)

(2)

where K(i) represents the proportion of the result generated by HSI threshold segmentation in super-pixel i. (5) Initialize a set proportion rate (rate should be between 0 and 1), then when the proportion of rape flowers to be segmented in the super-pixel is greater than the rate value, the super-pixel is kept as object. For this purpose, we set a reserved array R [1 … n], when K(i) > rate, set R(i) = 1, otherwise R(i) = 0.

24 Segmentation of Oilseed Rape Flowers Based …

231

After this operation, matrix S is transformed into binary image (1 represents the target to be segmented, i.e., rape flowers). And then it is converted to a color image and is displayed in Fig. 24.1d.

24.3 Experimental Results In this paper, the experimental object is oilseed rape in large area, and the acquisition method is aerial photography by UAV. The size of original images is 5472 × 3078, and the experimental image is 10% size of the original image. The segmentation experiment is simulated by MATLAB r2017b. The segmentation of rape image is combined with HSI color space threshold segmentation and super-pixel segmentation. We selected two groups of experimental results for analysis. In the following two sets of results, k = 1500 for super-pixel segmentation and rate = 0.5 for reservation. The experimental results are shown in Figs. 24.2 and 24.3. By observing the above two experimental results, we can find that the result of HSI threshold segmentation consists of obvious holes and a large number of backgrounds close to the target color. In the result of the combination of super-pixel segmentation

Fig. 24.2 a Original image; b HSI threshold result; c SLIC super-pixel result; d Result of fusion of the two

232

J. Zeng et al.

Fig. 24.3 a Original image; b HSI threshold result; c SLIC super-pixel result; d Result of fusion of the two

and HSI threshold segmentation, the background impurity has disappeared, most of the holes are completely filled, and the expected effect is achieved.

References 1. Fu, D., Jiang, L., Mason, A.S., et al.: Research progress and strategies for multifunctional rapeseed: a case study of China. J. Integr. Agric. 15(8), 1673–1684 (2016) 2. Aggelopoulou, A.D., Bochtis, D., Fountas, S.: Yield prediction in apple orchards based on image processing. Precis. Agric. 12(3), 448–456 (2011) 3. Thorp, K.R., Dierig, D.A.: Color image segmentation approach to monitor flowering in lesquerella. Ind. Crops Prod. 34(1), 1150–1159 (2011) 4. Shuai, D., Liu, C., Wu, X., et al.: Image segmentation of field rape based on template matching and K-means clustering. In: IOP Conference Series: Materials Science and Engineering. IOP Publishing vol. 466, no. 1, p. 012118 (2018) 5. Long, Y., Liu, C. H., Shuai, D., et al.: Image segmentation of canola based on color similarity in color space. In: Proceedings of the 2nd International Conference on Computer Science and Application Engineering. ACM, p. 123 (2018) 6. Achanta, R., Shaji, A., Smith, K., et al.: SLIC superpixels compared to state-of-the-art superpixel methods. IEEE Trans. Pattern Anal. Mach. Intell. 34(11), 2274–2282 (2012)

Chapter 25

Agricultural Remote Monitoring and Alarm System Based on LoRa and Internet of Things Cloud Platform TingTing Chen, XinChen Zhang, and BoWen Zhao

Abstract This system simplifies the solution of large-scale agricultural plant network application. This solution can help enterprise and individual developers connect their products to the Internet without massive network coverage. The system adopts stm32F4 series chips, integrating LoRa wireless transmission and cloud platform technology. The system can realize the visualization of monitoring results and the function of results warning. The advantages of this system make it possible for LoRa wireless communication technology to build large-scale self-organizing networks in the future.

25.1 Introduction At present, a variety of intelligent hardware has gradually entered people’s vision. However, in the existing Internet of things solutions, there are problems of high application development threshold and high platform coupling. This will greatly increase the cost of infrastructure equipment and remote monitoring [1]. Cloud computing combines distributed computing with the Internet, enabling IT services to be better served as resources for users. This makes it easier to combine small personal networks, medium community enterprise networks, and cloud services on large world networks [2]. At present, many companies at home and abroad have launched application platforms suitable for the Internet of things. This makes it easier for companies and individuals to connect their products to the Internet, greatly reducing the cost of using and developing large systems. T. Chen · X. Zhang (B) · B. Zhao Central China Normal University, Wuhan, China e-mail: [email protected] T. Chen e-mail: [email protected] B. Zhao e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_25

233

234

T. Chen et al.

Fig. 25.1 Schematic diagram of system design

25.2 System Implementation Method 25.2.1 System Design Requirements The system should have the characteristics of wide communication network coverage and random collection. But if you need to upload data to the network, it will require extensive network coverage. This is not possible in many rural areas and other underdeveloped areas [3]. LoRa technology can increase the communication distance between nodes and reduce the number of relay nodes deployed. This will provide data communication services for a star network of acquisition nodes and sink nodes. In the complex environment of remote areas, LoRa communication distance can reach more than 2 km. The system can also transmit the collected data to the cloud platform, which can monitor and remind users of their products [4]. The schematic diagram of system design is shown in Fig. 25.1.

25.2.2 Platform Introduction and Access Method OneNet cloud platform is a message-oriented middleware service for the Internet of things industry. At its core is the OneNet device service, which connects traditional control acquisition devices and terminal display devices to the HTTP protocol. The encryption part verifies the access password through the shared link access to ensure the security of the data, which can help users easily recover access rights. When incompatible data occurs, the information is sent to the device using the GET method of the HTTP protocol interface.

25 Agricultural Remote Monitoring and Alarm System Based …

235

25.2.3 Hardware System Design 25.2.3.1

Hardware Structure Model of the System

Data transmission and processing unit can complete sensor, mobile client, cloud platform with three kinds of data transmission and processing [5]. The data processing unit is connected to the main control board through UART to process the received data. The SPI bus sends the processed valid data to the single-chip microcomputer or the data transmission unit in time [6]. The data transfer unit is connected to the cloud platform or mobile client via a gateway. Data uploaded locally by the sensor can be transferred to the cloud platform in time, and data sent by the mobile terminal or cloud platform can also be sent to the processing unit.

25.2.3.2

Hardware Structure and Composition of the System

• Data collection section DHT11 digital wet-temperature sensor adopts single bus data format. Data can be divided into decimal parts and integer parts. A full data transfer is 40 bits, where the higher order takes precedence. Sensor data outputs uncoded binary data. Data (humidity, temperature, integer, decimal) should be processed separately. • LoRa module is electrically connected to MCU/ARM equipment The system adopts pass-through data mode. Transparent transmission refers to the communication between devices of the same address and the same communication channel [7]. User data can be character or hexadecimal. The schematic diagram is shown in Fig. 25.2.

Fig. 25.2 LoRa communication and transport mode

236

25.2.3.3

T. Chen et al.

LoRa Technical Features

In order to realize the low-power consumption and remote communication demand of the device, Semtech released the LoRa technology chip in 2013. The technology continues the low-power consumption characteristic of frequency shift keying modulation. At present, the technology can achieve longer communication distance and higher sensitivity. At the same time, as long as it is determined that the spread spectrum codes of different channel receiving devices are different, the data finally received will not cause crosstalk problem. • LoRa transmission time in the open area LoRa technology mainly adopts star network structure. In order to reduce power consumption, most LoRa structures use wake up method to collect data when the acquisition node appears. The mode of awakening includes active awakening and air awakening. In the wake-up mode, the time consumption is mainly the air transmission time of the packet, including the transmission time of the lead code and the transmission time of the header payload. • Data package structure The transmitted packets include the lead, header, CRC, and load-CRC. Modem includes display mode and implicit mode. The default value for the leading code is 12 characters. If the amount of data received is large, the length of the leading code can be shortened appropriately. During communication, the receiver periodically compares the length of the data transmitted by the sender with the length of its own local data. If the length is the same, it starts receiving data. In the case of inconsistencies, it will continue to stay asleep. The structure is shown in Table 25.1. • Wan data upload part WiFi module USES LVTTL to communicate with MCU (or other serial devices), and built-in TCP/IP protocol stack to realize the conversion between serial port and WiFi. If the STA mode is selected, the device will connect to the Internet through the router, and achieve seamless switching between LAN and WAN through the Internet control device. Table 25.1 Packet structure for transmission Lead code

The header Display mode

CRC

The payload

Load-CRC

25 Agricultural Remote Monitoring and Alarm System Based …

237

25.3 Communication and Encryption 25.3.1 Communication Protocol The control system protocol is mainly composed of three parts: HTTP protocol between data transmission board and cloud platform, serial port protocol between WiFi and master control module, and UART protocol between LoRa and master control module. Due to space constraints, the following is only a detailed description of the HTTP protocol between the data transfer board and the cloud platform. The cloud computing platform sends relevant information to the application in the form of HTTP/HTTPS requests. When using the push service, OneNET acts as the client, while the user’s third-party application acts as the server. Third-party applications need to support URL authentication and data reception.

25.3.2 Main Encryption Algorithm Due to the data transmission and wireless communication in the system, we encrypted the transmission of LoRa. Common public key cryptography is ECC, ECDH, or ECDSA [8]. The last two algorithms are based on ECC. The following is the conic expression (1). 

(x, y) ∈ R 2 , y 2 = x 3 + ax + b, 4a 3 + 27b2 = 0



(1)

RSA algorithm is the most widely used asymmetric encryption and decryption algorithm on the Internet, widely used in various fields of computer network security. RSA algorithm has the characteristics of simple mathematical principle and easy to implement. RSA algorithm is easier to crack than ECC algorithm. According to the key length decoding time test of RSA and ECC [9] algorithm, ECC algorithm is more difficult to decode and more complex to crack.

25.4 System Function Coordination System coordination plays an important role in the construction of Internet of things system. During debugging, different sensors, communication modules, and WAN access are used to debug the same MCU. Data processed by the hardware system is encapsulated and sent over HTTP. The cloud platform can receive and parse the set temperature value, and shows the operation status and alarm according to the received data.

238

T. Chen et al.

Fig. 25.3 Data receiver and test scenario

25.4.1 Data Receiving and Sending DHT11 digital temperature and humidity sensor collects data source. The two parameters communicate through a serial port. Device baud rates, data bits, and parity bits need to be adjusted before use. The data sink and test scenario are shown in Fig. 25.3. Set parameters: u8*temp(temperature), u8*humi(humidity). Temperature range: 0–50 °C, humidity range: 20–90%.

25.4.2 Cloud Platform Display and Alarm Since the cloud platform is primarily used for enterprise data storage and management, only the cloud receiving data is tested here. Device information needs to be added to the LoRa and WiFi functional frameworks, including device ID, APIKEY, data ID, and WAN-related private information. Two data streams (humidity and temperature) are set in the platform, and the detection results of the cloud platform are shown in Fig. 25.4.

25.5 Conclusions In this paper, the research status, development trend, and related communication network technology of modern portable temperature and humidity monitoring system are analyzed in detail. On this basis, the system adopts the new low-power LoRa wireless communication technology, which has the advantages of long communication distance. The mobile acquisition equipment has the advantages of low-power consumption, strong anti-interference, and high-cost performance. These characteristics will play an important role in information management and testing in the agricultural field. Of course, the system introduced in this paper still has many shortcomings that need to be improved. It is believed that in the future, the emergence

25 Agricultural Remote Monitoring and Alarm System Based …

239

Fig. 25.4 Part of the data received by the cloud platform

of Internet of things technology will completely change the traditional sensor application technology, which will be more widely applied in agricultural application scenarios or more fields without wireless network coverage.

References 1. Xiaohong, P., Xufeng, X., Hongjun, Z.: Intelligent aquaculture control system based on onenet interest of things cloud platform. Modern Comput. (Professional Edition). (31), 65–70 (2017) 2. Pike, R., Dorward, S., Griesemer, R., et al.: Interpreting the data: parallel analysis with Sawzall. Dyn. Grids Worldwide Comput. 13(04), 277–298x (2005) 3. Bo, Z., ZhongHua, D., Shuang, X., Wei, T.: Model constructing method for analyzing the trusty of cloud. J. Softw. 27(6), 1349–1365 (2016) 4. Chen, K., Zheng, W.: Cloud computing: system instances and current research. J. Softw. 20(5), 1337–1348 (2009) 5. Li, L., Zhang, Z.: Wireless transient temperature test system based on STM32. Instrum. Tech. Sensor (3), 96–97 (2014) 6. Junjie, Y., Xinchen, Z.: Smart water meter system based on lora wireless communication. Comput. Meas. Control 26(5), 288–291 (2018) 7. Zhengfeng, H., Lan, L.: The research on designing and optimizing of the algorithm for elliptic curve cryptography (ECC). ACTA Electronica Sinica 32(11), 1904–1906 (2004) 8. Aifen, S., Yixian, Y., Xinxin, N.: Research on the authenticated key agreement protocol based on elliptic curve cryptography. J. Beijing Univ. Posts Telecommun. 27(3), 28–32 (2004) 9. Honglin, Z., Xiaoqin, L., et al.: Remote control system of air conditioning system based on internet of things cloud platform. Comput. Eng. Des. 38(1), 265–270 (2017)

Chapter 26

Aerial Triangulation Study of Unmanned Aerial Vehicles in Forest Area Based on Pix4D Jing Zhao, Jian Guan, Yan-jie Li, Hong-wei Du, Zhi-min Zhu, Yue Zhao, and Teng-fei Ma Abstract This paper takes the Moon Island area of Changsha City in Hunan Province of China as the research object, and uses Pix4D mapper software to obtain data acquisition, control point measurement, aerial triangulation, and DOM data production and editing of UAV data in Moon Island area. The Pix4D mapper software-based solution for the whole process of the aerial triangulation of the drone is provided, which provides favorable data and technical support for aerial survey data processing.

26.1 Introduction The production process of photogrammetry 4D products mainly includes aerial triangulation, DLG, DOM, DEM, and DRG production [1]. The first process of air triangulation is to analyze the aerial triangulation. The aerial triangulation is the process of solving the ground space coordinates and the out-of-plane orientation elements of J. Zhao (B) · J. Guan · Y. Li · Z. Zhu Liaoning Ecological Engineering Vocational College, Shenyang, China e-mail: [email protected] J. Guan e-mail: [email protected] Y. Li e-mail: [email protected] Z. Zhu e-mail: [email protected] H. Du Jinzhou Medical University, Jingzhou, China Y. Zhao Medical College of Jinzhou Medical University, Jingzhou, China T. Ma Shenyang Dian Wei Information Technology Co. Ltd, Shenyang, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7_26

241

242

J. Zhao et al.

all the encryption points in the study area by the adjustment of the photogrammetry. At present, the commonly used aerial triangulation softwares include Inpho, PIX4D, and MapMatrix, which both support aerial triangulation of large aircraft data and drone data, and have special UAV modules for aerial triangulation of drone data [2]. Unmanned aircraft data with unstable flight attitude has obvious advantages. Among them, PIX4D has obvious advantages of fast processing speed and high precision. The accuracy is slightly lower than that of the other two softwares, but it can meet the drawing requirements of the forestry industry [3]. Team members use PIX4D software for air triangulation data processing to provide favorable data and technical support for aerial survey data processing.

26.2 Study Area and Data 26.2.1 Overview of the Study Area This paper takes the Moon Island forest area in Changsha City, Hunan Province of China as the research object. The area of the study area is 0.901 km2 . The forest coverage rate is high, and it is divided into forests, farmland, and residential areas. The project utilizes the Dajiang UAV for data acquisition, and then performs ground control point measurement and point recording production to complete the flight control work of the field.

26.2.2 Data Collection The materials that need to be prepared in advance include drone image data, POS data, camera files, and field control point data. First, check the integrity and quality of all data to eliminate unacceptable images. Check the POS data, mainly check the photo number of the change of the navigation belt to prevent the photo number in the POS data from corresponding to the photo data photo number. If there is no correspondence, you should manually adjust it [4]. The data format of the control point file: in order to facilitate the internal control point, the control point name contains a photo number where the point is located.

26.3 Process of Aerial Triangulation 26.3.1 Create a Project and Import Data Start the PIX4D mapper, set the project name in [Project]-[New Project]: yld4D, and select [Next]. Load image and Set POS data. According to the source of POS data,

26 Aerial Triangulation Study of Unmanned Aerial Vehicles …

243

Fig. 26.1 POS data attribute

set the precision of POS data (with horizontal accuracy of 5 m and vertical accuracy of 5 m) [5] (Fig. 26.1).

26.3.2 Quick Processing Check The results of fast processing are less accurate, so the speed of processing is much faster. Therefore, the quick processing proposal is carried out on the flight site, and the problem is found to be convenient and timely [6]. If the fast processing fails, the subsequent operations may have the same result, even if the subsequent processing is successful, the obtained result is not high precision.

26.3.3 Add Control Points and Puncture Points Control points must be reasonably distributed within the area of the survey, usually with control points around and in the middle of the survey area. There must be at least three control points to complete the reconstruction of the model [7]. Usually 100 photos are around 6 control points, and more control points will not have a significant improvement in accuracy (more control points can improve elevation accuracy where elevation changes are large). Do not make the control point too close to the edge of the measurement area. The control point can be found at the same time on five images (at least two). The control point data is displayed in the map window (shown in blue crosshairs). At this time, it is necessary to check whether the control point is in the flight area. If

244

J. Zhao et al.

Fig. 26.2 Control point map view

the control point is no longer in the flight area, the imported POS coordinate system and the coordinate system of the control point are inconsistent [8] (Fig. 26.2). Click on the view in the menu bar to open the empty three-ray editor. You can see the generated connection point and the predicted position of the control point (blue circle with a small dot in the middle), click the control point 881, and the attribute on the right. The predicted position of the control point 881 above the five images will appear in the column (shown in blue circle). Accurately penetrate the position of control point 881 in all the following images, and the remaining image puncture methods are the same as above [9] (Fig. 26.3).

26.3.4 Fully Automatic Processing Click the menu bar to run, select local processing, the system displays the following dialog box. This process requires control points to participate in the empty three calculations, generate reports and DSM DOM and other data, which takes a long time, about 5 h [10].

26 Aerial Triangulation Study of Unmanned Aerial Vehicles …

245

Fig. 26.3 Sting point completion

26.4 Result Analysis 26.4.1 Regional Cyberspace Error The regional cyberspace error is shown in the Fig. 26.4. The mean reprojection error is the error in the space, in pixels. The pixel size on the camera sensor is typically 6 microns (µm), which may vary from camera to camera. The unit converted to physical length is 0.166577 × 6 µm.

Fig. 26.4 Mean reprojection error

246

J. Zhao et al.

Fig. 26.5 Camera self-calibration error

26.4.2 Camera Self-calibration Error The upper and lower parameters can’t be too different (for example, if focal length is above 33.838 mm, the following is 20 mm, then it is definitely the initial camera parameter setting problem). R1, R2, R3 are the three parameters which cannot be greater than 1, otherwise serious distortion may occur (Fig. 26.5).

26.4.3 Control Point Error Error X, Error Y, and Error Z are errors in three directions. At the same time, at the end of the accuracy report, you can show which photos have been stabbed in the control points and which photos are not stabbed. If the accuracy is not good enough, you can puncture these points in these photos as needed to improve accuracy (Fig. 26.6).

26.5 DOM Edit Use the Mosaic Editor to edit the DOM and wait for a while when you first load it. After the DOM is loaded, the middle map window is displayed. It is possible to check how the house and other objects appear to be pulled and deformed, indicating that it is necessary to edit the area. In the DOM diagram, the deformation is more obvious in the corners [11]. Click “Draw” in the Edit Mosaic dialog box. In the image, click the left mouse button along the road mouse to click on the deformed building area, and the right mouse button ends. Select the first one in the replaceable image that appears on the right, click the left mouse button, and the image is automatically replaced in the DOM [12]. The method of building the deformation of the other areas is the same as described above. Repeat the above steps to complete the editing of the DOM image. Click

26 Aerial Triangulation Study of Unmanned Aerial Vehicles …

247

Fig. 26.6 Control point error

“Save” in the Export tab of the Edit Mosaic dialog box on the right to complete the DOM modified data save, and then click “Export” to complete the DOM data export to the specified path and replace the original initial DOM data. The exported DOM product is shown in Fig. 26.7.

26.6 Conclusion (1) After the efforts of the team members, the data acquisition of the drones in the study area and the layout of the field control points were completed. Improve powerful data support for aerial triangulation. (2) The team members used the Pix4D software to perform empty processing on the study area, and the horizontal and elevation accuracy reached 0.03 m, which met the accuracy requirements. (3) Complete the editing and modification of the DOM product, and generate the DOM result map of the research area. After team research, it has formed a solution for the entire process of air three processing, providing method guidance for the aerial survey data processing of other enterprises and college personnel.

248

J. Zhao et al.

Fig. 26.7 DOM map of research area

References 1. Shen, X.-H.: Opinion on the current work fine breed base in Liaoning Liaoning forestry. Sci. Technol. 01, 29–31 (2004) 2. Rajora, O.P.: Challenges and opportunities for conservation of forest genetic resources. Euphytica Springer J. 4, 136–137 (2001) 3. Maaten, T.: About management and research of forest genetic resources. For. Studies Metsanduslikud Uurimused 2, 56–57 (2010) 4. Hu, C.: Study on GIS-based planning and layout of Metasequoia glyptostroboides in Hubei Province. J. Northwest For. Univ. 02, 70–72 (2014) 5. Xiao-jie, W.A.N.G.: Study on tree species recognition and plant location method based on large scale aerials. J. Northeast For. Univ. 05, 60–65 (2005) 6. Li, J.-y., Liu, D.-d., Rui-Lv: Old bald sub-national nature reserve scientific research report. Shenyang Press, pp 11–65 (2016) 7. Wang, Wei: J. Beijing For. Univ. 04, 23–25 (2015) 8. Ryzhkova, V.: A gis-based mapping and estimation the current forest landscape state and dynamics. J. Landscape Ecol. 4, 34 (2011) 9. Julia: Factors affecting the accuracy of forest clear-cut area estimation on medium spatial resolution satellite winter images. For. Studies Metsanduslikud Uurimused 5, 122–124 (2008) 10. Zhao, J., Liu, D.-l., et al.: Application of GIS to study on multi-function operation in Jingouling forest farm. J. Northwest For. Univ. 25, 207–209 (2010) 11. Stere, K.: UAV and GIS based tool for collection and propagation of seeds material-first results. Remote Sens. Spat. Inf. Sci. 07, 213 (2016) 12. Liu, D.-l., Zhao, J.: The precise location research of the good breeding base–taking the old bald nature reserve as an example. Liaoning For. Sci. Technol. 1, 29–31 (2004)

Author Index

A Abe, Jair Minoro, 3, 11, 19 Aharari, Ari, 3, 19 An, Yi, 217

Heng, Zhou, 147, 157 Hou, Xueshi, 77 Huang, Bo, 119 Huang, Tianming, 201

C Chen, Jie, 31 Chen, TingTing, 233 Cui, Ningning, 191

J Jiang, Yuting, 47, 65, 77 Jiao, Yuan, 217

D Deng, Liangyi, 47, 77 Dong, Yan Hui, 89 Du, Hong-wei, 241

F Fang, Yong, 65, 77, 99 Fan, Ying, 129 Fu, Jianhang, 191

G Gou, Wan Li, 89 Guan, Jian, 241 Guo, Hui, 109 Guo, MingHao, 65 Guo, Ting Ting, 47 Guo, Yong, 99

H Han, Wei, 191 He, Chenying, 167

L Lan, Peng, 55 Li, Jiangli, 139 Lima de, Luiz Antônio, 11 Lin, Kaiyan, 31 Liu, Baiqiu, 77, 99 Liu, Chang, 31 Liu, Feng, 119 Liu, Gaoxiu, 183 Liu, Pingzeng, 191 Li, Xiujuan, 209 Li, Yan-jie, 241 Li, Yue, 217

M Mao, Tengyue, 119, 129 Ma, Teng-fei, 241 Mei, Han, 77

N Nakamatsu, Kazumi, 11, 19

© Springer Nature Singapore Pte Ltd. 2021 K. Nakamatsu et al. (eds.), New Developments of IT, IoT and ICT applied to Agriculture, Smart Innovation, Systems and Technologies 183, https://doi.org/10.1007/978-981-15-5073-7

249

250 O Oliveira de, Cristina Corrêa, 19 S Sheng, Haoxuan, 99 Si, Huiping, 31 Souza de, Jonatas Santos, 11 Souza de, Taciana Tamyris Alves, 19 Sun, Fenggang, 55 Sun, Kaiqiong, 227 T Tian, Jianyan, 139 W Wang, Jin Yi, 89 Wang, Liuqiang, 119 Wang, Wei, 217 Wang, Xianglong, 191 Wang, Xuan, 227 Wang, Yangyang, 99 Wang, Zhijun, 55 Wei, Hengbin, 129 Wei, Xue, 77 Wen, Li, 47 Wu, Bing, 47, 65 Wu, Junhui, 31 X Xiao, Wang, 147, 157

Author Index Xie, Yingjun, 183 Xu, Lin, 191 Xu, Zhenfeng, 209

Y Yang, Chunsheng, 3 Yang, Juan, 129 Yang, Lu Hua, 89 Yang, Xingchun, 177 Yifan, Zheng, 147, 157 Yi, Liu, 147, 157 Yin, Jingjing, 209 Yin, Wu, 167 Yong, Cao, 147, 157 Yuan, Zehui, 109 Yu, Yuting, 191

Z Zeng, Jiahua, 227 Zhang, Baohua, 217 Zhang, Chengfang, 177 Zhang, Jianyong, 191 Zhang, Sunan, 139 Zhang, XinChen, 233 Zhao, BoWen, 233 Zhao, Jing, 241 Zhao, Yue, 241 Zhao, Zhilong, 47, 65, 77, 99 Zhong, Xiaoling, 47, 65, 77, 99 Zhou, Shaoping, 109 Zhu, Zhi-min, 241