A forecasting method of forest pests based on the rough set and PSO-BP neural network

Abstract

In order to improve the forecasting accuracy of the occurrence period of insect pests, this paper proposes a kind of forecasting method based on the combination of rough set theory and improved PSO-BP neural network. It takes insect pests of Euphrates poplar forests as the object of study. First, an attribute reduction algorithm of rough set is used to eliminate redundancy attributes. Input factors of the forecasting model of insect pests (temperature, humidity and rainfall) can be reduced from sixteen to eight. Then, particle swarm optimization (PSO) algorithm is improved using the inertia weight, and weights and thresholds of BP neural network are optimized using the improved PSO algorithm. Finally, the forecasting model of insect pests is established using rough set and an improved PSO-BP network. The test results show that rough set theory can effectively reduce the feature dimension and the improved PSO algorithm can reduce the iteration times, with an average accuracy of 94.8 %. This method can provide the technical support for the prevention and control of the insect pests of the Euphrates poplar forests.