An improved double-vector model predictive control to eliminate multilevel jumps for 5L-ANPC
摘要
Double-vector model predictive control (DV-MPC) can effectively enhance the control performance of single-phase five-level active neutral-point-clamped (5L-ANPC) converters. However, existing switching sequences for DV-MPC of single-phase 5L-ANPC present inherent limitations in addressing critical issues such as multilevel jumps and DC-link capacitor voltage control. To resolve these problems, a DV-MPC method based on optimized switching sequences is proposed. It enables adaptive optimization of control objectives according to different system voltage levels. For low-voltage applications, priority is given to improving the dynamic response of DC-link capacitor voltage control. By selecting voltage vectors with superior capacitor voltage balancing capability, the balancing speed of DC-link capacitor voltages is significantly accelerated. For high-voltage scenarios, the mitigation of multilevel jumps becomes the primary consideration. Through refined switching sequences, smooth transitions between adjacent output levels are ensured, thereby eliminating all multilevel jumps that occur in conventional MPC implementations. Both simulation and experimental results are used to validate the superiority of the proposed method. In low-voltage applications, compared with traditional MPC, the dynamic response speed of DC-link capacitor voltage is improved by 76.04%, and multilevel jumps per output voltage cycle are reduced from 3 to 1. In high-voltage scenarios, the multilevel jumps are completely eliminated while achieving a 69.79% improvement in DC-link capacitor voltage balancing speed.