The Research On Forest Pest Forecast Based On Fuzzy Neural Network

Forest pest is one of the important forest disasters. Since it does not only cause the huge economic losses to China's forestry, but also restrict the sustainable development of the ecological environment severely, the people pay more and more attention to the prevention and control of forest pest. The forest pest forecast is the foundation and the important part of the prevention and control of forest pest. It's so-called "forewarned is profits, without prejudging the waste". The accurate prediction of forest pest can reduce the loss of forest resources and control the occurrence of disasters effectively.First, this article elicits the conception of the fuzzy neural network by introducing and comparing the related theory of the artificial neural network and the fuzzy technology, and summarizes the development and the classification of the fuzzy neural network and many varieties of the fuzzy neuron model. Then though combining, the design is proposed about the forest pest forecast based on fuzzy neural network. It constitutes the model structure of the RBF fuzzy neural network based on the T-S model and the BP fuzzy neural network based on the fuzzy clustering by the research of the fuzzy inference system, RBF neural network, BP neural network and learning algorithm.This paper utilizes the stepwise regression method to sift the meteorological factors which affecting the insect strain rate of Dendrolimus punctatus as the representatives of the forest pest, and determines the forecasting factors finally. It uses the two kinds of fuzzy neural network model structure to design the forest pest forecast through making the forecasting factors of Dendrolimus punctatus as the input and the insect strain rate of Dendrolimus punctatus as output. This design explains detailedly the each layer function of the two kinds of fuzzy neural network, the data flow and the corresponding algorithm design. Finally, this application design of the forest pest forecast is simulated by training. It establishes the multiple regression equations, the RBF neural network and BP neural network as contrastive experiment, and compares the prediction methods of the three different methods. The analysis by comparing shows that the prediction based on fuzzy neural network forecasting method is the more than multiple regression forecasting method and corresponding neural network forecasting method. The method based on fuzzy neural network makes average error rate lower and satisfactory forecasting.In summary, the forest pest forecast based on fuzzy neural network is not only can improve the accuracy of prediction, but also has extensive application value.