On the development of an optimal parametric fuzzy controller by genetic algorithms
Chih-Hsun Chou · 2002
One way for a systematic approach to fuzzy controller design is by applying genetic algorithms (GAs). GAs, however, are much more applicable to numerical-type optimization problems. A traditional fuzzy controller contains both linguistic-type rules and numeric-type reasoning. Hence, transforming a fuzzy controller design into a GA-applicable optimization problem becomes the first subject in the design approach. So, in this paper, we present an index function to represent the linguistic control rules in terms of numeric indices. In this way, a GA design approach becomes feasible. The index function has a tunable parameter which is adaptive to the controlled system and is novel to the fuzzy rule in a TSK (Tagaki-Sugeno-Kang) type fuzzy controller. Simulation results with a second-order damping system are presented to show the performance of the proposed fuzzy controller.