Model Predictive Control (MPC) is widely employed in three-phase two-level voltage source inverters (2L-VSIs) due to its fast dynamic response and straightforward implementation. However, conventional MPC requires the evaluation of all eight candidate voltage vectors (VVs), which increases computational burden and current prediction time, introduces higher common-mode voltage (CMV), and may degrade steady-state performance. To address these limitations, this paper investigates the effect of reducing the number of candidate VVs on CMV suppression, the reduction in current prediction time, and the enhancement of 2L-VSI performance. First, a five-voltage-vectors MPC approach is developed, achieving noticeable CMV suppression compared with the conventional approach. Although this approach achieved CMV suppression, it still incurred a high computational burden. Therefore, it was further developed into a systematic switching approach based on three VVs, in which only three candidate VVs are selected at each sampling instant. The proposed approach achieves two primary objectives: suppressing CMV and reducing the current prediction time by 50%. Experimental validation is conducted to compare the proposed approach with the conventional MPC in terms of CMV, current prediction time, Total Harmonic Distortion (THD), inductance variation sensitivity, dynamic response, and power loss. The results demonstrate that the proposed approach achieves superior steady-state and dynamic performance while significantly reducing the current prediction time and achieving suppression of the CMV at Vdc/2, thereby enhancing the performance of 2L-VSIs.
Medekhel et al. (2026) studied this question.