An Improved Computationally Efficient Multi-Vector Model Predictive Current Control for SPMSM Drives

Abstract

Although dual-vector model predictive current control (MPCC) provides fast dynamic response and a compact control structure, it still relies exclusively on the inverter’s eight basic voltage vectors (VVs). This dependence leads to substantial computational load as well as pronounced current ripple, ultimately limiting steady-state performance. To overcome these limitations, this paper presents an enhanced MPCC strategy for surface-mounted PMSM drives. First, a simplified optimal-vector selection framework is introduced to reduce the computational complexity of conventional MPCC, achieving nearly a 50% reduction in execution time. Second, a refined cost function incorporating both soft and hard constraints on current variation is developed to suppress rapid current changes and achieve near zero-error tracking. Finally, a duty-ratio allocation method based on the deadbeat principle is derived, enabling more effective utilization of the selected voltage vectors and further improving control performance. The proposed method is validated through theoretical analysis and simulation studies and is experimentally demonstrated on a 1-kW PMSM platform, confirming its effectiveness and superior steady-state and dynamic performance.