Optimal Model Predictive Torque Control for Three-Level NPC-Fed PMSM Drives

Permanent Magnet Synchronous Motors (PMSMs), recognized for their high-power density and excellent dynamic performance, are widely employed in emerging applications such as new energy vehicles and advanced manufacturing systems. To meet the increasing demand for high efficiency and low harmonic distortion, the three-level neutral-point-clamped (3L-NPC) inverter has emerged as a mainstream topology in multilevel drive systems, owing to its reduced voltage stress and superior output waveform quality. This paper presents an optimal model predictive torque control (MPTC) strategy for the PMSM drive system based on a 3L-NPC inverter topology. The strategy incorporates a simplification of the number of selectable voltage vectors and a DC-link capacitor voltage balancing mechanism. To reduce the computational burden in MPTC, a two-layer vector selection method is proposed to minimize the number of candidate voltage vectors. Additionally, a zero-vector duty ratio time allocation strategy is introduced to address the DC-link voltage balancing issue in three-level inverters, effectively suppressing voltage imbalances. The proposed strategy is validated through simulation results.