A Virtual Space Vector-Based Model Predictive Control for Inherent DC-Link Voltage Balancing of Three-Level T-Type Converters

Abstract

The finite-control-set model predictive control (FCS-MPC) has been widely studied to control voltage source converters (VSCs) due to its fast-dynamic response and robustness. To design an FCS-MPC for a three-level VSC, it is necessary to design a multiobjective cost function, which usually consists of a term dedicated to control the neutral-point voltage (NP-V) of the VSC dc link. Nevertheless, selecting weighting factors for the multiple control objectives is time consuming. To address this issue, this article proposes a new FCS-MPC, which only uses voltage vectors and virtual voltage vectors that do not affect the NP-V, such that an inherent dc-link voltage balancing can be achieved. The proposed approach not only simplifies the controller design by using a simplified cost function, which significantly reduces the execution time of the MPC, but also eliminates the voltage and current sensors for the dc link. Moreover, the proposed MPC can improve the quality of VSC output waveforms. The simulation studies are carried out to verify the superiority of the proposed MPC over the existing ones. The effectiveness of the presented FCS-MPC is also experimentally validated on a T-type converter prototype.