Aperiodic Model Predictive Control via Perturbation Analysis

Abstract — In this paper, an enhanced event-based scheme for model predictive control (MPC) of constrained discrete-time systems with additive disturbances is investigated. The re/calculation of the MPC control law is triggered whenever an event depending on the error of the measured state with respect to the nominal state of the system occurs. Between the controller updates, the last computed control trajectory is applied to the system, in conjunction with a correction term. This term consists of a perturbation solution of the nominal system which itself depends on the aforementioned error. The overall framework yields less conservative results with respect to our previous work. The results are illustrated through a simulated example. I