Abstract
In this paper, a new validity criterion that determines the optimal decomposition and the minimum linear model bank employed to design a controller for a nonlinear system is presented. Based on the gap metric, the proposed validity criterion quantifies the degree of compactness for a model-base. For a decomposition given a low degree of compactness, the optimal linear model bank that provides sufficient information for multimodel controller design of the nonlinear system is obtained without model redundancy. To illustrate the usefulness of the proposed method, multimodel H? control algorithm is applied to three nonlinear processes for set point tracking.
