FLC design for multi-objective system with varying weights

Abstract

For a system with multiple objectives, designing the rule base for the fuzzy logic controller(FLC) is not an easy task. In particular, when the fuzzy if-then rules are obtained from dictations of human operators, the rules can be inconsistent due to possibly conflicting multiple objectives. In this situation, we propose two types of FLC design schemes to satisfy all the objectives. The first design scheme is that by using varying weights according to the results of evaluating the outputs of the system. The proposed controller consists of a supervisory controller and many subcontrollers. Each subcontroller is a usual type of FLC responsible for each control objective and the supervisory controller coordinates the subcontrollers by adjusting the weights. The second design scheme is that based on a newly defined measure of inconsistency. For this purpose, the existing definition is critically examined and the measure of inconsistency suitable for human intuition is newly defined and then is utilized as weights for the rules to achieve the predetermined control performance. In addition, by means of simple examples, the proposed schemes are shown to be effective for multi-objective systems.