Citation format: Gui Zechun, Zhao Sijian. Review of the application of artificial intelligence in agricultural risk management[J]. Smart Agriculture, 2023, 5(1): 82-98. DOI：10.12133/j.smartag.SA202211004
GUI Zechun, ZHAO Sijian. Research application of artificial intelligence in agricultural risk management: A review[J]. Smart Agriculture, 2023, 5(1): 82-98. DOI：10.12133/j.smartag.SA202211004
Click CNKI to read
Click the official website to read (full text, free)
A review of the application of artificial intelligence in agricultural risk management
GUI Zechun, ZHAO Sijian*
(Institute of Agricultural Information, Chinese Academy of Agricultural Sciences, Beijing 100081, China)
Abstract: Agriculture is a basic industry related to the national economy and people's livelihood, but at the same time it is a weak industry, and traditional agricultural risk management research methods have problems such as insufficient nonlinear information mining, low accuracy and poor robustness. Artificial Intelligence (AI) has powerful features such as strong nonlinear fitting based on big data, end-to-end modeling, and feature self-learning to solve these problems. This paper first analyzes the research progress of AI in agricultural vulnerability assessment, agricultural risk prediction, and agricultural damage assessment, and concludes as follows:1. The feature importance assessment of AI in agricultural vulnerability assessment lacks scientific and effective verification indicators, and the application method makes it impossible to compare the advantages and disadvantages between multiple models, so it is recommended to use the subjective and objective method for evaluation. 2. In risk prediction, it is found that with the increase of prediction time, the prediction ability of machine learning models tends to decrease, overfitting is a common problem in risk prediction, and there are still few studies on the mining of spatial information of graph data. 3. The complex agricultural production environment and changeable application scenarios are important factors affecting the accuracy of damage assessment, and improving the feature extraction ability and robustness of deep learning models are the key and difficult problems that need to be overcome in future technology development. Then, corresponding solutions are proposed to the performance improvement problems and small sample problems in the process of AI application. For the performance improvement problem, according to the user's familiarity with artificial intelligence, a variety of model comparison methods, model combination methods and neural network structure optimization methods can be used to improve the performance of the model. For problems with small samples, it is often possible to combine data augmentation, generative adversarial networks, and transfer learning to enhance the robustness of the model and improve the accuracy of model recognition. Finally, the application of AI in agricultural risk management is prospected. In the future, the introduction of artificial intelligence into the construction of agricultural vulnerability curves can be considered; Aiming at the upstream and downstream relationships of agricultural industry chain and agriculture-related industry relationships, more graph neural networks are used to further study agricultural price risk prediction. In the process of damage assessment modeling, more expertise in the relevant fields of the assessment target can be introduced to enhance the feature learning of the target, and the expansion of small sample data is also the focus of future research.
Keywords: agricultural risk management; Artificial intelligence; vulnerability assessment; risk prediction; Damage assessment
Article image
Figure 1 Agricultural risk management cycle and corresponding analysis and evaluation
Fig. 1 Agricultural risk management cycle and corresponding analysis and evaluation
Figure 2 AI-based agricultural vulnerability assessment process
Fig. 2 Agricultural vulnerability assessment process based on artificial intelligence （AI）
Support Vector Machine (SVM), Recurrent Nerual Network (RNN), Long Short-Term Memory (LSTM), Graph Nerual Network (GNN), Graph Convolutional Network GCN）
Figure 3 AI-based agricultural risk prediction process
Fig. 3 Agricultural risk prediction process based on AI
Figure 4 AI-based agricultural damage assessment process
Fig. 4 Agricultural damage assessment process based on artificial intelligence
Figure 5 Neural network structure optimization
Fig. 5 Optimization of neural network structure
About the corresponding author
Sijian Zhao Researcher
Zhao Sijian, male, Ph.D., researcher, visiting scholar of The Hong Kong Polytechnic University, visiting associate professor of Central University of Finance and Economics, vice chairman and secretary general of the Agricultural Insurance Branch of the China Agricultural Risk Management Research Association, standing director of the Risk Analysis Professional Committee of the China Disaster Defense Association, and deputy director of the Agricultural Risk Management Research Center of the Institute of Agricultural Information, Chinese Academy of Agricultural Sciences. He has been engaged in agricultural risk management and agricultural insurance research for a long time, and has presided over and participated in more than 40 projects such as the National Natural Science Foundation of China Youth Fund Project, International Cooperation Projects and General Projects, Key Projects of the Ministry of Education, Science and Technology Support Projects of the Eleventh and Twelfth Five-Year Five-Year Science and Technology Projects of the Ministry of Science and Technology, Beijing Science and Technology Plan Project, Science and Technology Innovation Project of the Chinese Academy of Agricultural Sciences, Basic Scientific Research Business Funds Project of the Chinese Academy of Agricultural Sciences, etc., presided over and participated in the Ministry of Finance, the Ministry of Agriculture and Rural Affairs, the China Banking and Insurance Regulatory Commission and other national ministries and commissions, Pacific Property Insurance Company, Ping An Property Insurance Company, More than 20 projects commissioned by insurance institutions such as China Coal Property & Casualty Insurance Company, China Reinsurance Corporation, BOC Insurance, China Agricultural Reinsurance Company, etc., led the research and development of multiple agricultural insurance industry application systems and platforms such as China's agricultural production risk assessment and zoning map system, Beijing agricultural risk management and insurance information management platform, agricultural weather index insurance Internet service platform, etc., and has published more than 60 academic papers, including 8 in SCI, 20 in EI, and 15 in CPCI-S. The first applicant was granted 2 patents and 14 software copyrights; He has edited 1 monograph, 3 conference proceedings, participated in the editing of 1 monograph and 1 textbook, and has accumulated rich experience in agricultural risk management and agricultural insurance technology.
Source: Smart Agriculture (Chinese and English), Issue 1, 2023
Please contact the editorial office for reprinting
Supported units for this issue
Weichai Lewo Intelligent Agriculture Technology Co., Ltd
Zibo Institute of Digital Agriculture and Rural Research
Shanghai Zanqi Culture Technology Co., Ltd
small
shop
Welcome to shop
Smart agriculture WeChat exchange service group
In order to facilitate the academic exchanges of readers, authors and reviewers in the field of agricultural sciences, promote the development of smart agriculture, and better serve the majority of readers, authors and reviewers, the editorial department has established a WeChat communication service group, where issues related to discussion and submission can be consulted in the group. How to join the group: add Xiaobian WeChat 331760296, remarks: name, unit, research direction, Xiaobian pulls you into the group, institutional marketing and advertising personnel do not disturb.
Posting a call for submissions
Welcome to publish information such as scientific research team introduction, innovative scientific research achievements and related activities on our public account.