Model Predictive Current Control Based on Virtual Voltage Vector Method for Parallel Three-Level Inverters

Abstract

The parallel three-level neutral-point clamp (3L-NPC) inverters are widely solicited today to satisfy the higher power demand by using low-rated switching devices in industrial applications such as renewable power generation. With the 3L-NPC in parallel, the zero-sequence circulating current (ZSCC) occurs in the closed circuit and decreases the performances of the current response if the two 3L-NPCs share the same dc bus with no galvanic isolation at the ac side. From the analysis of the ZSCC equivalent circuit, the virtual voltage vectors are proposed and exploited by the finite control set model predictive control (FCS-MPC) for mitigating the ZSCC and improving the current accuracy. In order to keep the neutral-point balance and reduce the switching loss, the differential capacitor voltage and the average switching frequency are considered in the proposed MPC. An experimental platform based on the Simulink-Real-Time system is presented for further verification of the effectiveness of the proposed method. The simulation and the experimental results obtained under the proposed strategy show that the ZSCC is reduced to near zero, and the current response provided has lower THD.