Research on Fuzzy Neural Networks Model and its Application in Predicting the Precipitation State

The self-adaptation and self-studying features of neural networks have been combined with the logical reasoning ability of fuzzy system to produce a fuzzy system based neural networks, BP arithmetic is used to adjust system parameters. Finally the fuzzy neural networks model with the monthly mean rainfall is established, as compared with the model of multi-variant linear regression, the fuzzy neural networks is smaller in error values of forecasting, therefore the FNN is suitable to extend in long range forecast of rainfall and is of practical value.