A general scheme for multi-model controller using trust

Fuzzy logic theory, introduced by Zadeh (1965), has been used in different approaches in control theory. Fuzzy inference using a data base of an expert knowledge is the most widely used. A second approach, first presented by Sugeno, consists in the use of several models representing a partial view of a complex system and using fuzzy logic interpolation property to mix them. The present paper deals with a new concept in fuzzy inference mechanism where fuzzy logic is combined with trust notion derived from the possibility theory. This fuzzy inference mechanism has been used to control a nonlinear first-order process from the knowledge of linear first-order models and their associated optimal controllers. Computer simulations are presented to show the interest of this method in comparison with Sugeno multi-model controller.