Tuning Fuzzy Logic Controllers by Genetic Algorithms

The performance of a fuzzy logic controller depends on its control rules and membership functions. Hence, it is very important to adjust these parameters to the process to be controlled. A method is presented for tuning fuzzy control rules by genetic algorithms to make the fuzzy logic control systems behave as closely as possible to the operator or expert behavior in a control process. The tuning method fits the membership functions of the fuzzy rules given by the experts with the inference system and the defuzzification strategy selected, obtaining high-performance membership

functions by minimizing an error function defined using a set of evaluation input-output data. Experimental results show the method's good performance. KEYWORDS: fuz~ logic control systems, tuning, genetic algorithms 1. INTRODUCTION Recently fuzzy control techniques have been applied to many industrial processes. Fuzzy logic controllers (FLCs) are rule-based systems which are useful in the context of complex ill-defined processes, especially those which can be controlled by a skilled human operator without knowledge of their underlying dynamics. The essential part of the FLC system is a set of fuzzy control rules (FCRs) related by means of a fuzzy implication and the compositional rule of inference. Address correspondence to Francisco Herrera, Dept. of Computer Science and A.L, ETS de Ingenieda Inform6tica, University of Granada, 18071 Granada, Spain. *This research has been supported under project PB92-0933 Received July 1993; accepted November 1994. International Journal of Approximate Reasoning 1995; 12:299-315 © 1995 Elsevier Science Inc. 0888-613X/95/$9.50 655 Avenue of the Americas, New York, NY 10010 SSDI 0888-613X(94)00033-Y 300 F. Herrera, M. Lozano, and J. L. Verdegay FCRs are usually formulated in linguistic terms, in the form of IF-THEN rules, and there are different modes for deriving them [1]. In all cases, the correct choice of the membership functions of the linguistic label set plays an essential role in the performance of an FLC, it being difficult to represent the experts' knowledge perfectly by linguistic control rules. The fuzzy-control-rule base has many parameters, and its control de- pends on the tuning of the control system. Therefore, an FLC contains a number of sets of parameters that can be altered to modify the controller performance. They are [2]: • the scaling factors for each variable, • the fuzzy set representing the meaning of linguistic values, • the IF-THE N rules. Each of these sets of parameters have been used as controller parameters to be adapted in different adaptive FLCs. In this paper we present an adaptive FLC that modifies the fuzzy set definitions (it alters the shapes of the fuzzy sets defining the meaning of linguistic values) to determine the membership functions that produce maximum FLC performance according to the inference system (fuzzy implication and compositional operator) and the defuzzification strategies used--that is, to tune the FCR so as to make the FLC behave as closely as possible to the operator or expert behavior. This method relies on having a set of training data against...

Website: 150.214.190.154 | Filesize: -
No of Page(s): 17
Download Tuning Fuzzy Logic Controllers by Genetic Algorithms*.pdf