A Model Predictive Direct Current Control Strategy With Predictive References for MV Grid-Connected Converters With $LCL$-Filters

In this paper, a new model predictive direct current control (MPDCC) strategy is proposed for a medium-voltage (MV) neutral-point-clamped converter with an LCL-filter-based connection to the grid. The proposed strategy addresses the issues of resonance damping and harmonic attenuation through the virtual resistor (VR) concept, and is thus, termed MPDCC-VR. Because MPDCC is capable of achieving very long prediction horizons, the VR-based reference terms are predicted in conjunction with the state trajectories at each time-step, thus enabling more accurate decisions to be made by the controller. Simulation results verify the applicability of the MPDCC-VR to an MV case study. It is shown that the proposed approach exhibits a very good level of steady-state performance in the presence of grid voltage distortion, and at the chosen operating point is capable of substantially outperforming multiloop control with space vector modulation. Experimental results, which are provided for a down-scaled prototype, show excellent agreement with those obtained in simulation, further validating the proposed MPDCC-VR strategy.