An effective approach for optimal design of fuzzy controllers

The novel approach described in this paper represents a substantial departure from the complexity and consequent difficulties involved in the optimal design of fuzzy controllers. By reducing the number of design parameters of a fuzzy controller, the search in a hyperspace of 94 dimensions will collapse into a search in a space of 7 dimensions. In this way, a significant enhancement can be achieved in terms of mathematical simplification, time and efficiency. The proposed approach can still control a given system with a performance index very close to that obtained by full optimization. In this paper, the conventional design of fuzzy controllers is first reviewed. The problems associated with such design are then addressed. A novel view to the decision table allows the designer to come up with an efficient coding approach for the fuzzy controller. Based on such efficient coding, a genetic algorithm is then employed to find the free parameters of the decision table. To examine the efficiency of the proposed approach, an optimal fuzzy controller is designed based on this technique and the results are compared with those obtained by full optimization as well as conventional approaches.