Future Trends in Model Predictive Control

Abstract

Model predictive control is a control technique in which a finite horizon optimal control problem is solved at each sampling instant to obtain the control input. The measured state is used as initial state and only the first control of the calculated optimal sequence of controls is applied to the plant. A key advantage of this form of control consists in its ability to cope with complex systems and hard constraints on controls and states. This resulted in a wide range of applications in industry, most of them in the petro-chemical branch. In this survey, a selected history of model predictive control is presented, with the purpose to outline the principles of this control methodology and to analyze the progress that has been made. The initial predictive control algorithms, mainly based on input/output models, are recalled in the introduction and then we focus on the more recent work done in nonlinear model and hybrid model predictive control. The stability problem and the computational aspects are discussed to formulate some fruitful ideas for the future research.