Improved Model Predictive Control for Minimizing DC-Link Capacitor Current in Parallel Dual-Inverter-Fed PMSM Drives

Reducing the root-mean-square (RMS) current through DC-link capacitors is critically challenging in parallel dual-inverter-fed drive systems, as DC-link currents are heavily influenced by circulating currents and operating points. To address the complication, an improved finite control set model predictive control (FCS-MPC) scheme is proposed in this work. Unlike existing research, the proposed scheme accounts for circulating currents and power-sharing online when estimating DC-link ripple currents, ensuring consistent performance across a wide range of operating conditions. Moreover, the evaluation of DC-link performance is conducted over an extended prediction horizon, providing a more accurate representation of DC-link ripple current fluctuations around its average. To alleviate the computational demands imposed by a large number of switching states, a multi-rate hybrid structure is introduced. Within this structure, stator currents tracking can remain unaffected by other objectives, maintaining superior torque and speed response. Furthermore, DC-link currents prediction is performed within flexible subsets, keeping the calculation burden modest and allowing for real time application. Ultimately, the proposed scheme can achieve a 13.5% to 41.3% reduction in DC-link RMS current compared to the state-of-the-art approach.