Pseudo feedforward dynamic matrix control algorithm for desired closed-loop performance in MPC systems

Abstract

As a practical approach, the dynamic matrix control (DMC) algorithm has been widely used in the control of chemical processes for decades. Nevertheless, there is no strong relationship between the tuning parameters of DMC and the closed-loop dynamics. In this article, a pseudo feedforward dynamic matrix control (PFFDMC) algorithm for realizing desired closed-loop performance in industrial systems is proposed. Building on insights from cascade control, a pseudo feedforward compensator is added in front of the process to precondition the overall system offline for expected characteristics achievement. Then, the actual manipulated variables can be obtained first by solving a modified objective function and then through matrix transformation. The proposed algorithm provides a quantitative relation between the pseudo feedforward sequence and the final setpoint tracking results rather than a qualitative way. Numerical simulations on SISO and MIMO systems are performed to verify the effectiveness of the proposed algorithm on performance criteria achievement.