NONLINEAR DYNAMIC SYSTEM IDENTIFICATION BASED ON MULTI-MODEL WAVELET NETWORKS

Due to the highly nonlinear and complex dynamics of industrial system, it's difficult to build exact model. A multi-model wavelet network method to identify nonlinear dynamic system was presented. The structure and the learning method for the identification was given. We compared the performance of multi model wavelet network with that of single wavelet network. The simulation results showed that the method presented possessed high approximation accuracy and stability.