An Extended Linear MPC for Nonlinear Processes

Nonlinear behavior is a common feature of all real-world systems. However, for the sake of simplicity, a linear model is often used in the controller design procedure. Nevertheless, the neglected nonlinear dynamics could degrade the performance of controller drastically. This study presents a new method of designing a model predictive controller (MPC) for a class of nonlinear systems. In the proposed method, an MPC is first designed in state space based on a linear model and then modified by using modal series to compensate for the effect of the neglected nonlinear dynamics in the linear model. Because the proposed controller adjusts a linear controller instead of designing a new one, it can be easily applied in industries to modify controllers that have been designed based on linear models. In addition, its computational burden is much less than that of nonlinear MPC methods. In this study, the proposed technique is used to control two real-world systems, and the results of its application are discussed.