New Concepts for the Identification of Dynamic Takagi-Sugeno Fuzzy Models

Abstract

Takagi-Sugeno fuzzy models have proved to be a powerful tool for the identification of nonlinear dynamic systems. Recent publications have addressed the problems of local versus global accuracy and the identifiability and interpretability of local models as true linearisations. The latter issue particularly concerns off-equilibrium models. Well-established solution approaches involve techniques like regularisation and multi-objective optimisation. In view of a practical application of these models by inexperienced users this paper addresses the following issues: 1) unbiased estimation of local model parameters in the presence of input- and output noise. At the same time the dominance of the trend term in off-equilibrium models is balanced. 2) The concept of stationary constraints is introduced. They help to significantly improve the accuracy of equilibrium models during steady-state phases. A simulation model demonstrates the capabilities of the proposed concepts