A Tuning Method for the Membership Functions of a Fuzzy Controller

퍼지제어기의 멤버쉽함수의 튜닝 방법

  • Lee, Ji-Hong (Dept. of cont. Eng., Kumoh Nat'l Inst. of Tech. Univ.) ;
  • Chae, Seog (Dept. of cont. Eng., Kumoh Nat'l Inst. of Tech. Univ.) ;
  • Oh, Young-Seok (Dept. of cont. Eng., Kumoh Nat'l Inst. of Tech. Univ.)
  • 이지홍 (금오공과대학 전자계측공학과) ;
  • 채석 (금오공과대학 전자계측공학과) ;
  • 오영석 (금오공과대학 전자계측공학과)
  • Published : 1993.04.01

Abstract

It is known that the performance of a fuzzy controller is related with fuzzification method, inference rules, defuzzification method, and linguistic variables. Among these, generally, the linguistic variables and control rules are transfered to control engineers from an expert or experts of the controlled system and other parts are designed by control engineers. However, there may be some missed infirmations or uncertainties in the transfered data. The purpose of the paper is to propose an algorithm to tune the membership functions of initially given fuzzy sets To do so, a simple shape of the membership fuction is assumed for the fuzzy sets, and the relations between the shapes of the fuzzy sets and the performance of the control system is derived. According to the relations, the shape of the membership functions are modified during operation of the whole system. The proposed algorithm will be applied to two emample plants, type 1 and type 0 systems.

Keywords