Risk assessment method of agricultural products cross-border e-commerce logistics based on fuzzy neural network

As the basic necessity of human life, agricultural products have become the basic elements of economic and social development. Because of the complexity and uncertainty of the internal and external environment, the supply chain of agricultural products has various risks, such as vulnerability and instability in the system structure. In recent years, with the overall improvement of people's living standards, consumers demand for the market of fresh agricultural products has been expanded rapidly and food safety issues has also been a great deal of attention. Therefore, supply of fresh agricultural products with good quality is the main concern in traditional and e-commerce transportation. In this study, the traditional supplier evaluation methods and fuzzy neural network system were compared and analysed. Results of the current study explore that the fuzzy neural network method avoids the subjectivity of weight assignment in supplier evaluation problem, and fully reflects the mapping relationship of "evaluation index -evaluation conclusion", which is very persuasive. Keeping in view the aim of the evaluation process of cross-border agricultural products e-commerce logistics, the fuzzy neural network is designed and trained in detail, including sample pretreatment, node number determination, learning algorithm and learning rate selection, therefore, it has strong generalization ability. It is found that among the eight risk factors, the risk level of distribution factor, loading and unloading factor, environmental factor, storage and packaging is high, the risk level of information factor and cooperation factor is medium, and the risk level of customs and strategic culture factor is low. Based on the findings of the current study, the fuzzy neural network evaluation process of cross-border e-commerce logistics of agricultural products is proposed, which makes the evaluation more clear.