Martin Klaučo will give a seminar on "MPC-based Reference Governors: Theory and Applications" on Thursday, July 6, 2017 at 11:30 at IAM in room no. 641.
Abstract: This dissertation thesis deals with the improvement of the control performance of closed-loop systems via optimizing the setpoints for these closed-loops. Closed-loops consist of a primary controller, which provides stability and some tracking properties for a process. Since these controllers are usually of very simple structure, they do not provide constraint satisfaction nor enforce tracking properties. This thesis summarizes a concept of reference governors (RG) based on model predictive control (MPC), which provides optimal setpoints for the primary control layer. We will show how such an MPC-based reference governors (MPC-RG) are formulated. We will be dealing with three main cases, first, a closed-loop containing PID controllers, second a closed-loop utilizing the behavior of the on/off controller and finally a complex closed-loop scheme with local MPC controllers. We will show how to model these closed-loops and subsequently formulate the MPC-RG control problems, which are in the form of optimization problems. The second part of this thesis is devoted to case studies, where the benefits of the MPC-RG strategy is elaborated. We consider one experimental case study involving the stabilizing and control of a magnetically suspended ball. Next, we offer two simulation-based case studies, one focused on improving the behavior of a boiler-turbine system and the second concerning a thermostatically controlled temperature in buildings.