Dynamic Model Predictive Control

Abstract

In this paper an alternative approach to model predictive control is presented. In traditional MPC a finite horizon open loop optimal control problem is solved in each sampling instance. When uncertainties such as computational delays are present, one can encounter problems. We propose to parametrize the control sequence in each sampling instant in terms of a linear feedback controller, i.e. in each sample a dynamic feedback compensator is computed. Thus, if computational delays are present the control system runs in closed loop, decreasing the need for ad hoc solutions used in traditional MPC.