LAUSR

The modular multilevel matrix converter (M3C) is a promising topology for high-voltage high-power applications. Recent researches have proved its significant advantages for adjustable-speed motor drives compared with the back-to-back modular multilevel converter. However, the branch energy balancing in the M3C presents great challenge especially at critical-frequency points where the output frequency is close to zero or grid-side frequency. Generally, this balancing control depends on the appropriate injection of inner circulating currents and the common-mode voltage (CMV), whereas their values are hard to determine and optimize. In this paper, an optimization-based predictive control method is proposed to calculate the required circulating currents and the CMV. The proposed method features a broad frequency range balancing of capacitor voltages and no reactive power in the grid side. For operation at critical-frequency points, there is no increase on branch voltage stresses and limited increase on branch current stresses. A downscaled M3C system with 27 cells is designed and experiment results with the R–L load and induction motor load are presented to verify the proposed control method.