Cooperative Model Predictive Control Concepts for Coupled AC/DC-and DC/DC-Power Converters

Abstract

Model predictive control (MPC) concepts find an increasing application in the control of electronic power converters of different topologies. This is mainly due to the fast increase in computing power, e.g., by the application of system-on-chip architectures (FPGA <inline-formula> <tex-math notation="LaTeX">$+$</tex-math> </inline-formula> DSP on a single chip). MPC strategies are usually developed for a single power converter. In many real applications, the power converters are coupled to other power converters, where decentralized MPC strategies do not yield optimal control performance. On the other hand, centralized MPC concepts that control the overall coupled converter system are not practically feasible due to the increase in computational demands. For this reason, this work studies the utilization of a data link between the power converters to use cooperative MPC strategies. In contrast to earlier works on this topic, it is investigated how a long prediction horizon of recently developed MPC concepts can be beneficially used for the control of coupled power converters. In particular, cooperative MPC strategies are proposed that use the predicted system behavior transferred by the other coupled converters. The benefits of these strategies will be demonstrated by comparison to a decentralized and a centralized MPC strategy. It will be shown that the proposed strategy also allows to reduce the size of certain system components as, e.g., the dc-link capacitor, which entails reduced costs and an increased system dynamics.