Fuzzy rules acquisition using interactive genetic algorithms

Abstract

This paper discusses fuzzy rules acquisition and their linguistic expressions using the interactive GA. The number of fuzzy rules is also determined by the presented approach. A fuzzy rule is represented by a chromosome, and individuals with various numbers of chromosomes are considered by the GA operations. Fuzzy sets in the acquired fuzzy rules are expressed using plain linguistic labels. We attach great importance to a qualitative evaluation of rule intelligibility as well as a quantitative evaluation by a fitness function. The presented approach is applied to car control for running through crank-typed courses. Simulation results show that the acquired fuzzy rules approach is acceptable from the viewpoints of rule intelligibility, the number of fuzzy rules and control results.