A Terminal Set Feasibility Governor for Nonlinear Model Predictive Control

The terminal set feasibility governor (FG) is a model predictive control (MPC) add-on unit that guarantees the feasibility of the underlying optimal control problem by manipulating its target reference. Since the FG requires little information from the closed-loop system, it can be attached to any MPC law that tracks a piecewise constant reference. This paper extends the FG from linear to nonlinear systems as long as the nonlinear MPC features an invariant terminal set. The closed-loop FG+MPC satisfies constraints, is asymptotically stable, and exhibits finite-time convergence of the reference. Moreover, numerical simulations show that adding the FG significantly reduces the computational burden of the MPC for only a minor cost in performance.