Short horizon nonlinear model predictive control

This article concerns nonlinear model predictive control of the multivariable, open-loop stable processes whose delay-free part is minimum-phase. The control law is derived by using a discrete-time state-space formulation and the shortest “useful” prediction horizon for each controlled output. This derivation allows to establish the theoretical connections between the derived nonlinear model predictive control law and the discrete-time globally linearizing control, and to deduce the conditions for nominal closed-loop stability under the model predictive control law. Under the nonlinear model predictive controller, the closed-loop system is partially governed by the zero dynamics of the process, which is the nonlinear analog of placing a subset of closed-loop poles at the zeros of a process by a model algorithmic controller