Linguistically oriented fuzzy logic control and its design

Abstract The main advantage of the classical fuzzy controller (FLC) should be the ease of its design, which is close to the human way of thinking. However, tuning its performance requires modification of membership functions, and thus the result may be very far from the original linguistic description. In this paper, we analyze the standard Max-t-norm interpolation and compare it with logical inference. Several logical inference rules suitable for approximate reasoning are presented. We deal with the idea of a fuzzy controller that is both linguistic and logical in the highest possible degree (LFLC). It interprets if-then rules as linguistically expressed logical implications which are treated as special axioms in fuzzy logic. Therefore, it is more closely related to the human operator's language, which is “understood” by LFLC without the need to specify and modify the membership functions of fuzzy sets of inputs and outputs.