Selection Of Optimized Fuzzy Rules For Process Control Systems: Using Genetic Algorithms

Abstract

In this paper modeling, selection and optimization of fuzzy rules for multivariable feedback process control and decision making systems are developed and discussed. A general model for multivariable fuzzy rules is derived and sets of fuzzy parameters are assigned to the consequents of each rule. A genetic algorithm is used to optimize these parameters. A genetic algorithm search technique is also developed for the selection of the sub-set of appropriate rules from a larger theoretically possible set. The procedure is implemented on a multivariable dynamic model and results of extensive simulation studies are presented to demonstrate a satisfactory performance of the proposed approach