Fuzzy system identification using an adaptive learning rule with terminal attractors

Abstract

A fuzzy model with an adaptive learning rule is proposed to approximate a class of nonlinear systems. The proposed adaptive learning rule uses the concept of terminal attractors and guarantees that the identification is stable and fast convergence in finite time. Another subject of this paper is to determine the appropriate rule number of the fuzzy model required for a desired modeling error. A learning algorithm is also introduced to identify the centers of membership functions of fuzzy outputs.