Model Predictive Control for Five-Level Current Source Converter with DC Current Balancing Capability

Multilevel current source converter (MCSC) is an effective way to increase the power rating of conventional current source-fed motor drive, especially its current rating. However, there are still several challenges existing in this configuration, which mainly focus on its modulation and control techniques, and the inherent DC current imbalance issue. In this paper, a model predictive control (MPC) scheme with DC current balancing capability is proposed for a five-level current source converter (5L-CSC). The optimal space vectors for 5L-current source rectifier (5L-CSR) and current source inverter (CSI) are selected for their respective control objects. Moreover, the DC-side model of a 5L-CSC is analyzed, which is used for DC current prediction based on the applied switching states at both 5L-CSR and CSI sides. Finally, optimal switching states are selected and applied to realize DC current balancing. The proposed MPC scheme is verified on a (2MW/4160V/278A) back-to-back 5L-CSC system through simulation.