Edge Computing-Enhanced K-means and MobileNetV3 for Short-Term Photovoltaic Forecasting

This paper presents an edge computing-enhanced model integrating K-means clustering and MobileNetV3 for accurate short-term photovoltaic (PV) power forecasting. The proposed approach addresses challenges in data quality, model generalization, and real-time adjustments, improving forecast accuracy across various weather conditions. The study demonstrates significant improvements in forecasting metrics, such as RMSE and MAE, through multi-window correction factors, showcasing the model's robustness and adaptability in PV power prediction.