This study advances data-driven anomaly detection in PV systems, reviewing ML and change-point methods, proposing a hybrid detection model.