A multiple model approach to nonlinear internal model control design

Abstract

A nonlinear internal model control (NIMC) design using multiple model is proposed. In this design approach, the process model is described via validity functions as the weighted sum of a set of local models. Subsequently, an approximate inverse of the process model is constructed in order to design NIMC controller. It is shown that the validity functions can be treated as time-varying uncertainties in the resultant NIMC structure, and a structured singular value test for mixed time-varying and time-invariant perturbations is used to assess the performance limitation of the proposed NIMC design. Simulation results of a continuous stirred tank reactor illustrate that the proposed NIMC design method is superior to the conventional IMC design. © 1999 Elsevier Science Ltd.