Anomaly Detection in Solar PV Plants Using Drone Technology and AI for Proactive Maintenance

Solar photovoltaic (PV) power plants are growing rapidly, so advanced monitoring techniques are needed for their efficient operation and maintenance. Manual inspections are labour and time-intensive, making them unviable for larger installations or hard-to-reach remote locations. This work utilizes drone technology in conjunction with thermal imaging and machine learning algorithms for the early detection of failing PV modules. Such thermal discrepancies can come in the way of hotspots, potential induced degradation (PID), and open circuits and can be rapidly identified by drones equipped with infrared (IR) cameras that indicate potential defects in PV modules. Additionally, CNN and eXtreme Gradient Boosting (XGBoost) machine learning techniques further improve the accuracy of line defects classification. Extensive research supports the use of drone-based thermographic inspections, demonstrating that they enhance detection and reduce costs compared to manual monitoring. The proposed approach enables real-time fault detection, mitigates energy losses, and prolongs the lifespan of PV systems. This work contributes to the ongoing advancements in effective, automated PV plant surveillance for sustainable and economical solar power generation.