Design of robust fuzzy neural network controller with reduced rule base

Abstract

Listen

Translate article

Designing an effective learning algorithm to find the best rule and optimal structure is a major problem in building fuzzy controllers. This paper contributes a new alternative for the synthesis of Takagi-Sugeno fuzzy logic controller with reduced rule base. A hybrid learning algorithm called Hybrid Approach to Fuzzy Supervised Learning (HAFSL) which combines the Multiobjective Genetic Algorithms (MGA) and gradient descent technique is proposed for constructing a Robust Fuzzy Neural Network Controllers (RFNNC). This RFNNC is similar to nonlinear PI/PD controllers. Two phases of design and learning process are presented in this work. An MGA is used for finding near optimal structure/parameters of the fuzzy neural controller that minimises the number of rules (initialization procedure). The second stage of learning algorithm uses the backpropagation algorithm based on gradient descent method to fine tune some scaling factors, membership functions and the consequent parameters of the proposed controller. The genes of chromosome are arranged into two parts, the first part contains the control genes and the second part contains the genes parameters representing the fuzzy knowledge base. The effectiveness of this chromosome formulation enables the fuzzy sets and rules to be optimally reduced. Simulations demonstrate that the proposed robust control has successfully met the design specifications for efficiency and robustness, respectively.