A Fuzzy Wavelet Neural Network Model for System Identification

Abstract

In this paper, a fuzzy wavelet neural network model is proposed for system identification problems. The proposed model is obtained from the traditional Takagi-Sugeno-Kang (TSK) fuzzy system by replacing the consequent part of fuzzy rules with wavelet basis functions that have time-frequency localization properties. We use a radial function of Mexican Hat wavelet in the consequent part of each rule. A fast gradient algorithm based on quasi-Newton methods is used to obtain the optimal values for unknown parameters of the model. Simulation results of some benchmark problems in the literature are also given to illustrate the effectiveness of the model.