Locally Optimal Takagi-Sugeno Fuzzy Controllers

Abstract

This paper presents a new suboptimal design for stable fuzzy controller. Using fuzzy plant model and locally optimal gains (based on LQR design), we suggest a sufficient condition for stability of the closed-loop system that can be expressed as Linear Matrix Inequaloities. Provided that we have an accurate fuzzy model of the plant, the real system is globally stabel too. This new approach facilitates the design process much yielding sub-optimal global design. Simulation results show that even if we have not an exact fuzzsy model representation, this method increases attraction domain of the system comparing with conventional linear optimal contrllers.