On efficiency of identification methods for MIMO non linear system using Takagi Sugeno models

Abstract

AbstractThis paper deals with the identification of Takagi-Sugeno's (TS) models for non linear MIMO systems. The Takagi-Sugeno's model combines linear models with non linear membership functions. Two-stage approaches were proposed which estimate the parameters of the non linear and linear parts separately and recursively. The contribution of this article is to show the nonoptimality of the scheme when gradient techniques are used in the first stage. A more efficient algorithm using one-stage estimation is proposed.