Fuzzy control based on new type of Takagi-Sugeno fuzzy inference system

We suggested new type of fuzzy inference systems (FIS) based on Takagi-Sugeno. We called it enhanced fuzzy regression (EFR). New FIS uses fuzzy coefficients in right parts of the fuzzy rules. Fuzzy approximation theorem has been proved and learning procedure has been suggested for the EFR. We compared EFR with Mamdani FIS and concluded that EFR can be more effective for fuzzy control.

I. Introduction

Fuzzy logic has been progressed greatly since the first publication in 1965 [1]. Nowadays fuzzy logic and fuzzy inference systems (FIS) are widely used for fuzzy modeling [2], [3], [4] in such engineering applications as fuzzy control (including fuzzy PID) [5], [6], decision making [7], [8], and knowledge discovery from databases [9], [10]. Fuzzy control based on FIS has some advantages compared to classic control. Sometimes, the mathematical model of the control process may not exist, or may be too expensive in terms of computer processing power and memory. In such situation fuzzy systems based on fuzzy logic and empirical rules may be more effective. Some other advantages of fuzzy control are presented in [11]: •

Flexible, intuitive knowledge base design.

•

Convenient user interface.

•

Easy computation.

•

Easy learning.

•

Fuzzy logic is a “natural” way of expressing uncertain information.

•

Some fuzzy inference systems are universal approximators.