The rapid global expansion of photovoltaic (PV) power generation has increased the need for efficient operation and maintenance (O&M) strategies. One of the most significant contributors to PV underperformance is soiling—accumulation of dust, pollen, bird droppings, and other contaminants that reduce irradiance capture. Traditional manual cleaning is labour-intensive, costly, and often inefficient for large utility-scale solar farms. This research proposes an integrated system that combines Artificial Intelligence (AI) for electrical power system analysis with drone-based autonomous cleaning technology to optimize PV farm performance.
The proposed framework utilizes AI for soiling prediction, energy-loss estimation, anomaly detection, and optimal maintenance scheduling. Drones equipped with computer vision and autonomous cleaning payloads execute targeted cleaning only when economically beneficial. The system further integrates predictive generation modelling and power system state estimation to reflect cleaning effects on grid behaviour. Results from simulated and pilot-scale deployments indicate potential annual energy yield improvements of 8–16% and O&M cost reductions of up to 35%. The study demonstrates that combining AI-driven analytics with drone-based cleaning represents a promising pathway toward intelligent, sustainable, and cost-effective solar farm maintenance.