Two-Stage Model Predictive Control of Neutral-Point-Clamped Inverter-Fed Permanent-Magnet Synchronous Motor Drives Under Balanced and Unbalanced DC Links

Abstract

Controlling a three-level neutral-point-clamped (3L-NPC) inverter-fed permanent-magnet synchro-nous machine (PMSM) drive is a challenging task, especially when the dc-link voltage is unbalanced. The unbalanced dc-link voltages, which are with some of the voltage vectors varying in magnitude and phase, lead to highly distorted stator current. In this article, a two-stage model predictive control (MPC) is proposed for the 3L-NPC inverter-fed PMSM drive. In the proposed MPC method, control of the 3L-NPC inverter is decoupled into two stages. In the first stage, three switching states are determined by small hexagon selection using finite control set MPC. The rest switching states are controlled using the multiple vector MPC in the second stage. In the two-stage MPC, the number of switching state evaluations can be significantly reduced, and the nonlinearity caused by the unbalanced dc links can be included in the duty cycle optimization process, leading to improved output current quality under unbalanced dc links and reduced computational burden. Moreover, the output switching frequency is fixed due to the employment of the multiple vector MPC. A thorough experimental evaluation of the proposed MPC method with the deadbeat control with the pulsewidth modulation strategy has been conducted to validate the superiority of the proposed method.