Inductive Reasoning Supports the Design of Fuzzy Controllers

Issue title: Fuzzy Logic and Neural Networks: Theory and Applications

Article type: Research Article

Authors: Cellier, François E. | Mugica, Francisco

Affiliations: Department of Electrical and Computer Engineering, The University of Arizona, Tucson, AZ 85721 | Departament d'ESAII, Universitat Politècnica de Catalunya, Diagonal 647, 2da. planta, Barcelona 08028, Spain

Abstract: In this article, a new systematic design methodology for fuzzy controllers is presented. For any desired plant output, it is possible to find off-line the optimal plant input that will produce a plan output that is as close as possible to the desired one. However, this constitutes an open-loop design. In this article, a new methodology is introduced that allows computing a signal on-line that is close to the optimal plant input as a function of system inputs and plant outputs. To this end, an inductive reasoning model is created that estimates the optimal plant input from given system inputs and plant outputs. The inductive reasoning model can be interpreted and realized as a fuzzy controller. Thereby, a large portion of the controller is realized through feedback, and the previous open-loop design is converted to an equivalent and more robust closed-loop design. The inductive reasoning technique is described in detail in the first part of this article. An example is shown in the second part of the article to demonstrate the validity of the chosen approach.

DOI: 10.3233/IFS-1995-3107

Journal: Journal of Intelligent and Fuzzy Systems, vol. 3, no. 1, pp. 71-85, 1995