Explainable Incipient Fault Detection Systems for Photovoltaic Panels

Explainability improves maintenance decisions: The system uses XAI techniques (specifically LIME) to explain why faults are detected, helping field engineers and technicians understand the reasoning behind automated diagnoses rather than simply receiving a "fault detected" alert.

Hybrid approach reduces false alarms: By combining physics-based modeling (the irradiance-based three diode model) with machine learning (XGBoost), the system addresses a key challenge in data-driven fault detection—false alarms—which is critical for maintaining trust in automated monitoring systems.

Early fault detection matters: The system is specifically designed to detect "incipient faults" (early-stage problems) in photovoltaic panels, which is important for preventing equipment failures and optimizing maintenance scheduling before major breakdowns occur.

Transparency enables human oversight: The explainable AI application allows human experts to verify and understand the system's decisions, supporting responsible deployment where automated systems augment rather than replace human judgment in critical infrastructure maintenance.

Better performance in challenging conditions: The novel three-diode model improves accuracy at low irradiance conditions, demonstrating that responsible AI systems must be designed to work reliably across diverse real-world operating scenarios, not just ideal conditions.