Multi-voltage-vector-based modulated model predictive controller for three-level NPC inverters with neutral-point voltage balancing

The multi-level inverter (MLI) is more efficient due to its ability to eliminate low-order harmonics. Thus, it is widely used in renewable energy generation, electric vehicles (EVs), etc. The choice of the neutral-point clamped (NPC) converter topology as a research object is of great significance in practical engineering. However, complex modulation algorithms and neutral-point voltage unbalance are two inherent problems for such NPC inverters. In this paper, an improved multi-voltage vector model predictive control (MVV-MPC) strategy based on neighboring four voltage vector synthesis is presented for three-phase NPC inverters. The duty cycle calculation and derivation methods for the proposed method are very simple, and they were obtained through the action time of each voltage vector being inversely proportional to the value of the corresponding cost function. Moreover, the fixed switching frequency of this strategy is maintained and the current peak ripple is reduced, guaranteeing the preferred total harmonic distortion of the output waveform. Furthermore, a voltage balancing method based on zero-sequence voltage injection is utilized to achieve dynamic balancing of the midpoint voltage difference and to limit the oscillation magnitude of the midpoint voltage difference to within ± 1 V. Finally, MATLAB simulations and an HIL experimental platform are built to demonstrate the effectiveness of the proposed method.