Indirect Finite Control Set Model Predictive Control of Modular Multilevel Converters
Journal article, Peer reviewed
MetadataVis full innførsel
OriginalversjonIEEE Transactions on Smart Grid 2015, 6(3) 10.1109/TSG.2014.2377112
The modular multilevel converter (MMC) is a potential candidate for medium/high-power applications, specifically for high-voltage direct current transmission systems. One of the main challenges in the control of an MMC is to eliminate/ minimize the circulating currents while the capacitor voltages are maintained balanced. This paper proposes a control strategy for the MMC using finite control set model predictive control (FCS-MPC). A bilinear mathematical model of the MMC is derived and discretized to predict the states of the MMC one step ahead. Within each switching cycle, the best switching state of the MMC is selected based on evaluation and minimization of a defined cost function. The defined cost function is aimed at the elimination of the MMC circulating currents, regulating the arm voltages, and controlling the ac-side currents. To reduce the calculation burden of the MPC, the submodule (SM) capacitor voltage balancing controller based on the conventional sorting method is combined with the proposed FCS-MPC strategy. The proposed FCS-MPC strategy determines the number of inserted/bypassed SMs within each arm of the MMC while the sorting algorithm is used to keep the SM capacitor voltages balanced. Using this strategy, only the summation of SM capacitor voltages of each arm is required for control purposes, which simplifies the communication among the SMs and the central controller. This paper also introduces a modified switching strategy, which not only reduces the calculation burden of the FCS-MPC strategy even more, but also simplifies the SM capacitor voltage balancing algorithm. In addition, this strategy reduces the SM switching frequency and power losses by avoiding the unnecessary switching transitions. The performance of the proposed strategies for a 20-level MMC is evaluated based on the time-domain simulation studies.