A fixed rule number fuzzy controller with an adaptive degree of similarity

This paper proposes a controller that requires minimum knowledge of the process under study. The rule base is initially empty and is constructed on-line. The maximum number of rules is established à-priori (independently of the process order), so that one can have total control over the sampling time. For defuzzification an adaptive similarity factor is defined aiming to classify the fired rules in a qualitative manner. The controller presented was tested not only in simulations, but also in a real experimental process. It showed good results even in the presence of disturbances or process dynamic changes. This controller illustrates also the meaning of the Theory of Fuzzy Equality relations.