A data-driven droop control strategy for reactive power sharing and stability enhancement in islanded AC microgrids

In islanded microgrids, distributed generation (DG) units employing conventional droop control often exhibit significant mismatches in reactive power sharing. This issue primarily arises from the mismatches in line impedances. These inaccuracies not only disrupt reactive power balance but also pose potential risks to system stability. This study proposes an adaptive droop control strategy that integrates conventional droop control with the Estimation of Signal Parameters via Rotational Invariance Technique (ESPRIT) for online dominant mode identification, enhanced by Particle Swarm Optimization (PSO). ESPRIT is selected for its low computational burden and robustness, while PSO efficiently handles nonlinear optimization problems. The proposed controller dynamically adjusts the reactive power droop gain to achieve accurate reactive power sharing and simultaneously enhance the microgrid stability margin. Effectiveness of the proposed approach is evaluated on a 34-bus microgrid benchmark and compared with conventional droop, virtual impedance-based control, and modified reactive power-voltage (Q-V) droop control strategies. Simulation results demonstrate that the approach achieves precise reactive power sharing and improves the stability margin by 12.36 times compared to conventional droop control. The controller also maintains robust performance under communication link failures. Real-time validation using the OPAL-RT simulator confirms practical applicability. The proposed strategy provides a scalable, communication-efficient solution for stability and reactive power sharing, enhances understanding of the interaction between stability and power sharing, and supports reliable operation and renewable integration. The study is limited to linear loads and does not consider communication delays, which will be addressed in future work.