A new method to construct fuzzy systems based on rule selecting

The paper describes a new method to construct a fuzzy system for solving the problem how to select the number of the rules. The proposed method calculates the cluster center through the K-means algorithm but need not to presuppose the number of clusters which is determined according to the convergence of the cluster principle function being used. The convergence rate of the cluster principle function become slower when the number of the cluster is becoming bigger, and the proper number of clusters can be determined when that rate is slow enough to some degree. Then combining with recurrent least-squares algorithm to identify the parameter of the consequent part of the rule base, a fuzzy system is established. The simulation results demonstrate the efficiency and feasibility of the proposed method.