Nonlinear model predictive control: control and prediction horizon

摘要

In this paper a new class of Model Predictive control algorithms for nonlinear systems is introduced. A prediction horizon longer than the control horizon is used. It is hown that in this way it is possible to enlarge the domain of attraction and to improve the performance without enlarging the dimension of the optimization space. The idea is developed to solve the regulation problem and the tracking problem for reference signals constant beyond a prescribed future horizon.