Long-Horizon Direct Model Predictive Control for Medium-Voltage Converters Connected to a Distorted Grid

Abstract

Long-horizon finite control set model predictive control (FCS-MPC) is known for its superior performance, particularly when applied to complex, higher order systems, such as grid-connected converters with $LCL$ filters. This article proposes a long-horizon FCS-MPC method that effectively operates such systems even in the presence of time-varying model parameters and distorted grid voltage with variable harmonic content. To do so, the proposed method incorporates information about the grid voltage distortion when generating the reference trajectories of the controlled variables, namely, the grid and converter currents and the filter capacitor voltage. In addition, a fast estimation algorithm continuously updates the grid- and converter-side reactances, thus ensuring robustness to parameter variations in the system model. Real-time tests conducted in a hardware-in-the-loop environment validate the effectiveness of the proposed control approach across various operating conditions.