Weather Analysis to Predict Rice Pest Using Neural Network and D-S Evidential Theory

Agriculture, especially rice cultivation, has been challenged by various problems over the past few decades, with the problem of crop failures leading to crop failures being more particularly acute. Therefore, it is necessary to make predictions before the outbreak of pests, and take timely prevention and control measures to reduce the damage caused by pest outbreaks. This paper would select the farmland environmental data around Chongqing to study the rice pest prediction algorithm based on neural network and D-S evidential theory. Under the condition of small amount of environmental data, the relationship between farmland climate environment and pests is discussed. In this paper, BP neural network and Elman neural network are used to predict pests respectively. Then the neural network prediction results are used as weights. The combination decision ideas in D-S evidential theory are used to perform weight fusion, and new prediction results are obtained. The experimental results show that compared with the traditional prediction scheme of neural network, the prediction performance of the method combined with D-S evidential theory is better than any single neural network model, which can better reveal the relationship between climate factors and pest outbreaks.