Automatic rule generation for fuzzy controllers using genetic algorithms: a study on representation scheme and mutation rate

Abstract

A common difficulty in fuzzy systems is the need for their rules to be specified by a human designer. Following their successful application to a variety of learning and optimization problems, genetic algorithms (GAs) have been proposed as a learning method that enables automatic rule generation for fuzzy controllers. Fusion of fuzzy systems and genetic algorithms has recently attracted interest and a number of successful applications have been reported. However, there are some aspects to be considered when genetic algorithms are used for generating fuzzy control rules. In this paper, we discuss representation and mutation rate. We also attempt to find the representation scheme and mutation rate fit for automatic fuzzy rule generation when using GAs.