Fitting fuzzy membership functions using genetic algorithms

Using fuzzy logic to solve control problems has increased considerably. This makes the teaching of fuzzy control in engineering courses an urgent issue. Therefore, a self-training computer package in fuzzy control theory for students was developed previously. The package has all the necessary instructions for understanding all principles of fuzzy control. The training instructions are given through an application drill. Though this approach proved to be effective, by giving students a way of understanding an actual situation, it has a limitation: the learning method itself. The students always use the “trial-and-error” method to arrive to an adequate control action. The problem with this method is that students may be driven to the wrong conclusion that fuzzy control system corrections are but a matter of supposition. The purpose of the paper is to present a strategy for the automatic adjustment of membership functions using genetic algorithms