Performance Improvement of Model-Free Predictive Current Control for PWM Rectifiers Under Nonideal Power Grids

Abstract

This paper proposes an improved robust predictive current control strategy, combining an improved ultralocal model with a hybrid space vector modulation (HSVM) scheme. Compared to conventional ultralocal model, the proposed improved ultralocal incorporates grid voltage dynamics into the model architecture and replaces online estimations with direct sampled-value. The proposed control strategy enhances tracking accuracy during grid voltage fluctuations, especially under unbalanced and harmonically distorted conditions. Additionally, the HSVM strategy dynamically switches between three voltage vector sequences according to modulation index thresholds, which improves steady-state performance in non-ideal grid conditions. To address the prevalent issues of voltage fluctuations and harmonic distortion in practical power grids, existing solutions inject compensation into the complex power reference. However, existing methods overlook the negative-sequence current effects, which may lead to inaccurate results under unbalanced conditions. This paper incorporates the effects of negative-sequence currents and derives the power compensation term under non-ideal grid conditions, yielding an accurate analytical expression. Experimental results demonstrate that the proposed method exhibits strong robustness against inductance parameter variations and reduces current THD by 26% (ideal grid) and 11% (non-ideal grid) compared to conventional methods.