In recent years, taking advantage of neural networks, a lot of researchers focus their attention on the field forecasting. This paper introduced current situation of theories and applications of artificial neural networks in forecasting science. Exampled with the exploration of hydroelectricity in Sichuan, a detailed case of NN combinatorial forecasting is presented. This paper develops mathematical models respectively with BP Neural Nets, Radial Basis Function Nets and GMDH Nets, and forecasts electric energy demand in five possible patterns in the following twenty years with a combine of these three NN algorithms. Based on systems analyses of combinatorial forecasting results, the current surplus in electricity supply is temporary and relative. Electric power industrial should develop in advance and try to be a new pillar industry and a point of economy growth.
|