Piecewise Linear Steady–State Target Optimization for Control Systems with MPC: a Case Study

Abstract

This paper is concerned with predictive control combined with set–point optimization in the case of fast changing disturbances. The problem is encountered in many practical applications. Because of high computational complexity, nonlinear economic optimization cannot be repeated frequently. Therefore, in practice an additional steady–state target optimization repeated as often as the MPC (Model Predictive Control) algorithm is used. Typically, the steady–state target optimization is based on a linear steady–state process model. Unfortunately, in some cases, as the one studied in the paper, the target set– point optimization based on linear or linearized models fails. It is demonstrated in the paper that a solution to this problem can be the piecewise linear approximation of the nonlinear steady–state process model in the target optimization. The research is done for the control system of a MIMO chemical reactor. The presented results clearly show the effectiveness of the proposed approach.