Model Predicitve Control

Model predictive control (MPC) is a very successful optimization-based control technique. It consists of repeating the following process online: at each sampling time, solve a finite-horizon optimal control problem and apply the first part of the corresponding optimal solution to the system up to the next sampling time. The main advantages of MPC and reasons for its popularity and widespread success in many application areas are that (i) satisfaction of hard constraints on states and inputs can be guaranteed, (ii) optimization of some performance criterion can directly be incorporated in the controller design, and (iii) it can be applied to general nonlinear systems with possibly multiple inputs.

In our research, we focus on different aspects of current interest in MPC. This includes the development of robust, stochastic, and adaptive MPC schemes for uncertain systems, the consideration of more general control objectives than setpoint/trajectory tracking (economic MPC), the study of distributed MPC schemes for large-scale systems, as well as data- and learning-based MPC methods.