Adaptive fuzzy internal model control

Abstract

The Internal Model Control (IMC) structure is composed of the explicit model of the plant and a stable feed forward controller. The major task for the IMC design is to find a precise model of the plant. In this work Takagi-Sugeno fuzzy modeling is used to approximate nonlinear systems. Modeling by linearization is employed at certain operating points and separate local linear models are obtained. The Takagi-Sugeno fuzzy model is constructed to fuzzily switch among the local linear models. In order to obtain a more precise model, a normalized least mean square algorithm is added in parallel to the fuzzy model to adaptively quanta for the model mismatch. An adaptive inverse control concept using adaptive finite-impulse-response (FIR) filters has been used to implement the feedforward-controller part of the IMC structure.