Constrained finite receding horizon linear quadratic control

Abstract

Issues of feasibility, stability and performance are considered for a finite horizon formulation of receding horizon control for linear systems under mixed linear state and control constraints. It is shown that for a sufficiently long horizon, a receding horizon policy will remain feasible and result in stability, even when no end constraint is imposed. In addition, an efficient off-line computational scheme is presented which provides not only a stabilizing horizon length, but guaranteed performance bounds for the receding horizon policy. These calculations are demonstrated on both an open-loop stable and an open-loop unstable example.