GENETIC TUNING OF FUZZY CONTROLLER

Abstract

A methodology of applying Genetic Algorithm (GA) to optimize a fuzzy controller in a speed control using the integral of the square of the error (ISE) criterion is presented in this study. Motivated by the claim that fuzzy controller is hard to train, this study is conducted on the stretch of a computer simulation using MATLAB.

A cursory exploration has been done to determine the suitable combination of operators to be used in the optimization proper. Since GA is a random process, several trials were done in every exploration, as well as in the optimization proper.

Results showed that GA was able to tune the fuzzy controller. This derivative-free algorithm returned a lower average ISE than what is reported in Chuy's work, that is, using derivative-based optimization method, by 2.7 percent.