Feasibility governor for MPC with disturbance preview information

Abstract

This paper presents a feasibility governed preview MPC (PMPC) strategy for the control of constrained linear systems with previewed disturbances. The feasibility governor (FG) minimally modifies the supplied reference command so that the predicted terminal state of the PMPC policy remains in a robustly control invariant terminal set. As a result, recursively feasibility of the combined FG-PMPC strategy can be achieved with shorter prediction horizons than a standard PMPC strategy. Hence, the amount of computational effort and preview information necessary to implement PMPC is reduced. Conditions are established that guarantee finite-time convergence of the FG reference command and input-to-state stability of the closed-loop system. An adaptive cruise control example is reported to demonstrate a scenario where both preview information and the FG are necessary to ensure safety and real-time feasibility of the PMPC policy.