Flexible Predictive Direct Power Control for Distributed Generation Converters During Asymmetrical Grid Faults

The reliable operation of grid-connected distributed generation converters is challenged by severe unbalanced conditions and stringent fault ride-through requirements. To address these issues, this paper presents a sensorless flexible predictive direct power control (SF-PDPC) strategy for converters operating under severe asymmetrical grid faults. The proposed approach combines a frequency-adaptive neural network quadrature signal generator (FANN-QSG)-based virtual-flux estimator with a flexible power-reference generation scheme, enabling predictive control without grid-voltage sensors, conventional synchronization units, or cascaded filtering stages. The key feature of the proposed method lies in a flexible power-reference formulation that exploits the degrees of freedom associated with positive- and negative-sequence power components, allowing continuous regulation of the trade-off among current quality, active-power oscillations, and reactive-power oscillations under unbalanced grid conditions. This enables a unified control framework adaptable to different grid support objectives. The effectiveness of the proposed strategy is validated under a severe type-C voltage sag, grid frequency deviation, and harmonic distortion. Compared with the conventional PDPC, the proposed method reduces current total harmonic distortion from 57.78% to 0.44% while maintaining satisfactory active power tracking performance. Furthermore, the FANN-QSG-based estimator and the overall control structure demonstrate strong robustness under highly disturbed operating conditions. The proposed SF-PDPC enhances the operational flexibility of predictive power control for grid-connected converters operating under highly disturbed and unbalanced grid conditions.