The genetic algorithm based tuning method for symmetric membership functions of fuzzy logic control systems

A fuzzy logic control system which was tuned by different genetic coding schemes were examined. The first coding scheme is to encode the allowed shifting range between the adjacent centroids of membership functions. The second coding method is to encode the three parameters of a set of symmetric membership functions (SMPs). And the third one is to encode the rule table. The effectiveness of using these approaches were compared one another by controlling a linear time invariant second order plant. By using a general quadratic cost function, we analyzed step responses of varied coding strategies. These coding methods were evaluated by both the averaged dropping rate of the cost and the averaged elapse time per generation. Simulations showed the proposed encoding scheme which uses genetic algorithm searching on the best parameter set of symmetric membership functions outperforms the other two schemes in the delineated systems