Virtual Vector Based Model Predictive Control for Three-Level Sparse Neutral Point Clamped Inverter

The three-level sparse neutral point clamped (3L-SNPC) inverter is widely used for the system loss reduction and the neutral point voltage oscillation mitigation. The traditional model predictive control (MPC) only selected one voltage vector from 21 vectors, which resulted in the large current ripples. In order to improve the power quality, a virtual vector based model predictive control (VVMPC) strategy for 3L-SNPC inverter system is proposed in this paper. Firstly, the medium virtual vectors are constructed with two adjacent large vectors combing with the same dwell time, which increases the voltage vectors from 21 to 27. Then, the voltage vector reference is derived based on the mathematic model of 3L-SNPC inverter. Finally, two vectors are selected from 27 voltage vectors, and adopted to synthesize the voltage vector reference in real time to reduce the current ripples. In addition, a pair of redundant small vectors are adjusted reasonably to balance the neutral point voltage. With the proposed method, the improved performance of the system is realized, which is verified by simulation results.