Predictive control of NPC three-level inverter model based on current increment

Abstract

This paper proposes a Current Increment Model Predictive Control (CI-MPC) strategy for three-level inverters to address limitations in current tracking performance and computational burden associated with conventional methods. Unlike traditional Finite-Control-Set MPC (FCS-MPC) and Optimal Switch Sequence MPC (OSS-MPC), CI-MPC reformulates the control objective using current increments, significantly improving dynamic current tracking accuracy. The strategy incorporates a large-sector state transition matrix and upward vector rounding, reducing the number of required vector evaluations per control cycle to fewer than 2. This contrasts with 27 evaluations for FCS-MPC and 10 for OSS-MPC, substantially decreasing computational load. Simulation and experimental results demonstrate that CI-MPC outperforms FCS-MPC and OSS-MPC. Specifically, its dynamic current response speed is approximately 50 % faster than FCS-MPC and 80 % faster than OSS-MPC. The current total harmonic distortion (THD) under CI-MPC is reduced to 1.92 %, compared to 4.52 % for FCS-MPC and 3.01 % for OSS-MPC. Additionally, controller optimization time is reduced from 2.4µs (OSS-MPC) to 1.2µs, improving algorithm efficiency by approximately 50 %.