A new model of fuzzy neural networks and its application

We summarize fuzzy neural network´s research actuality, development tide and application field, expound basic conception of fuzzy logic system, artificial neural and fuzzy neural network; set up a normal fuzzy neural network model and study the algorithm aim at actual problem. The node number of fuzzy layer, normal layer and rule layer is computable if the model has an assured input and output pattern and fuzzy layer´s subject function, and the model has good applicability to export express and pattern identification. The combination model is applied to synthetic integration of forecast rainfall data produced by gradual regression method, periodic analysis plus multi-layer method and model output statistics method. The model is trained by short-term rainfall data of Zhejiang province from 1980 to 1997. The synthetic integration (forecast) results from 1998 to 2000 show that the presented model can obtain satisfactory forecast performance.