Stabilization of Constant Power Loads Using Model Predictive Control - Applied to Train Propulsion Systems

Industrial supervisors: Henrik Mosskull and Olle Trollberg, Bombardier

Abstract: This thesis considers stabilization of constant power loads (CPLs) fed by a dc power source through an input filter, using model predictive control (MPC). Train propulsion systems generally utilize electrical motors whose output torque is tightly regulated by power converters. Often, these systems behave as CPLs. When a CPL is coupled with an input filter it can lead to a stability problem known as the negative impedance instability problem. Current state of the art regulators deal with this problem using classical optimization-based controllers, such as . This thesis instead proposes a linear parameter-varying model predictive controller (LPV-MPC). This advanced control method solves the negative impedance instability problem while also being capable of explicitly addressing signal constraints, which often exist in power converter applications. The regulator is evaluated in MATLAB/Simulink as well as in a software-in-the-loop simulator. It has furthermore been realized in a real-time hardware-in-the-loop simulator and tested in a power laboratory. Theoretical results show improvements over controllers under certain operating conditions and the results can be used as a benchmark of theoretical performance limits for design of other regulators.