Adapting the gain of an FLC with genetic algorithms

Fuzzy logic controllers are knowledge-based systems. incorporating human knowledge into their knowledge base through fuzzy rules and fuzzy membership functions. The definition of these fuzzy rules and fuzzy membership functions is generally affected by subjective decisions, having a great influence over the performance of the fuzzy controller. In some cases, the membership functions are defined within a normalized interval, and the knowledge base includes a set of scaling functions to convert the input variables from its real value to a normalized one, and the normalized outputs to its real value. Different works have proposed the application of generic strategies. with a learning purpose, to the knowledge base of FLCs. The learning is usually centered on the membership functions and/or the rule base, using fixed (predefined) scaling functions. In this paper, the evolution is applied to modify the gain of the controller (by modifying the scaling function of each input or output variable) and the rule base. The use of linear and nonlinear scaling functions is analyzed. (C) 1997 Elsevier Science Inc.