An implementable receding horizon controller for stabilization of nonlinear systems

Abstract

The authors construct an implementable receding horizon controller which requires the optimal open-loop control to be updated only at a discrete set of times. This controller uses a relaxed terminal constraint x(T) in W (where W is some neighborhood of the origin). Since the terminal constraint is relaxed, the receding horizon strategy loses its stabilizing properties inside W. To compensate for this effect, a linear, locally stabilizing controller designed for the linearized system is used inside W. The resulting concatenated control action yields a controller which is shown to be globally stabilizing. The robustness properties of such feedback are analyzed.