A Hybrid Clustering Algorithm for Fuzzy Functions System Identification

A fuzzy function system identification algorithm based on hybrid clustering method is proposed on account of the problems of uncertain structure and parameter identification complexity of conventional fuzzy system.Compared with general fuzzy function systems,the new algorithm combines with the sample distance of fuzzy C-means and fuzzy C-regression model.In model prediction section,the membership degree of each input variable is computed by Gaussian function,and the input vector's membership degree is obtained by the fuzziness operator of input variable's membership degree.The algorithm is applied separately to Box-Jenkins gas furnace data,a double entry single output nonlinear system and Mackey-Glass chaotic time series.Experimental results show that this algorithm has good identification effect,which validates the effectiveness and practicability of the algorithm.