Enhancing dynamic operation optimization feasibility for constrained model predictive control systems

Abstract

AbstractThis paper addresses a novel model predictive control algorithm for tracking to achieve an optimal operation for discrete‐time linear systems. The soft constraints characterizing the desired system performance are adjusted to solve the infeasibility problems caused by the original constraints. In this method, based on the principle of steady‐state error and constraints violation minimization, we firstly optimize the transition steady‐state with refined terminal invariant set, and slack variables as well as consider the dynamic constraints of the state constraints. Two types of soft constraints and slack variables are introduced to adjust state constraints online so that the MPC is feasible. Then, a dynamic optimization problem is solved which employs the constraints with minimum slack variables and takes the resulted steady‐state as the temporary target in current control period. The proposed method is able to minimize the violation of constraints and provides more potential to improve the control performance. Besides, the recursive feasibility and closed‐loop stability can be guaranteed by the combination of the new soft constraints adjustment method and tracking scheme. Finally, simulations at different circumstances prove the efficiency of the proposed method.