A Learning System of Fuzzy Control Rules Based on Genetic Algorithms

This paper describes a fuzzy rule learning system called SLAVE (Structural Learning Algorithm in Vague Environment) which learns a set of fuzzy rules from a set of examples. SLAVE has been developed for working with noise-aaected systems where the application of some conditions of classical learning theory do not produce good descriptions. This learning system allows the structure of the rule to be obtained, i.e., it can determine those that are relevant for describing the system from all the variables proposed (feature selection). Finally, we test the behaviour of this learning algorithm in a control problem and we compare its results with other learning algorithms for control.