Dynamical multi-rule generation for self-organizing adaptive fuzzy controller

This paper presents a dynamical multi-rule generating algorithm for H_∞ adaptive fuzzy controllers. This algorithm aims at reducing the rule generating frequency. A high frequency of rule generating can cause undesired control force oscillations. The design object of this fuzzy rule-base management is to exploit the capacity of reducing rule generating frequency by a organized rule base and a dynamical multi-rule generating mechanism. This dynamical multi-rule generating mechanism reduces rule generating frequency by considering the prediction of system´s future motion. And, the organized rule base reduces rule generating frequency via enlarging the union of the effective area of each rule in the rule base. The other advantage of this rule base structure is that it can reduce the size of fuzzy rule base. The applicability and efficiency of this new algorithm are demonstrated by simulation of the invert pendulum example