Prediction-Based Interval Voltage Regulation Involving PV-BESS Coordination

Modern distribution networks face critical voltage security challenges due to high penetration of photovoltaic (PV) systems and battery energy storage systems (BESSs). Traditional regulation strategies have proven inadequate in mitigating rapid voltage violations caused by power fluctuations. Therefore, a prediction-based proportional coordinated power compensation method is proposed for interval voltage regulation (VR). Firstly, an improved convolutional neural network model is established to predict variable power flows by considering PV and load characteristics, enabling accurate voltage profile prediction. Then, the crucial bus with the most voltage violation is identified, and the optimal coordination ratios for reactive power compensation (RPC) and active power compensation (APC) are determined by an optimization process based on dynamic sensitivity analysis. Furthermore, the crucial bus VR method sequentially allocates RPC from PV inverters and APC from BESSs using a sensitivity-based proportional coordination. Finally, the test results demonstrate an effective VR across all buses over a 24-hour operation, with total RPC and APC requirements reduced by $41.8\%$ and $34.5\%$ in a real 10-bus network, and by $79.1\%$ and $39.5\%$ in the IEEE 123-bus network, respectively.