Fuzzy control has been used successfully in many practical applications. In traditional methods, experience and control knowledge of human experts are needed to design fuzzy controllers. However, it takes much time and cost to design fuzzy controllers.
In this thesis, an automatic design method for fuzzy controllers using genetic algorithms is proposed. The proposed method includes an effective encoding scheme, a new crossover and mutation operator. The maximum number of linguistic terms, which are described as input/ouput fuzzy set membership functions, is restricted under a certain number to reduce the number of combinatorial fuzzy rule search space. Those proposed crossover and mutation operators maintain the semantic relationship between membership functions and control rules. The desirable evolution characteristics of fuzzy controllers by the proposed method are shown by benchmark problems which are truck backing problem, cart centering problem and inverted pendulm problem. The result of experiments have been satisfiable compared with other design methods using genetic algorithms.