An Adaptive Model Predictive Control for Reduced Switch Count Three-Level Inverter

Abstract

The reduced switch count three-level neutral-point-clamped (RSC-3LNPC) inverter is becoming popular due to the advantage of cost reduction. However, maintaining the lower current ripple and balancing the dc-link voltage is always a challenging task, especially when the medium vector is missing. To overcome this limitation, this article proposes an adaptive model predictive control (MPC) with optimized vector and duty cycle to guarantee the lower current ripple and achieve the dc-link voltage balancing. First, the voltage vector reference is derived based on the mathematic model of RSC-3LNPC inverter. Then, the virtual medium vectors are constructed with two adjacent large vectors, which increases the voltage vectors from 21 to 27. On the basis of voltage vectors, two or three vectors are optimally selected to synthesize the voltage reference. Finally, the capacitor voltage balance is achieved by selecting redundant small vectors to avoid the weighting factor tuning process. The proposed MPC features a lower current ripple and improves the dynamic-state performance while guaranteeing the dc-link voltage balancing. With the proposed method, the performance of the system is improved, which is verified by simulation and experimental results.