Enhanced Fault Detection in Photovoltaic Panels Using CNN-Based Classification with PyQt5 Implementation

Area of Science:

• Renewable Energy Engineering
• Artificial Intelligence in Energy
• Materials Science for Photovoltaics

Background:

• Solar photovoltaic (PV) systems are crucial for renewable energy, but module end-of-life management is a growing concern.
• Regular maintenance and inspection are vital for PV system longevity, energy efficiency, and environmental protection.
• Detecting anomalies early can prevent significant performance degradation and system failures.

Purpose of the Study:

• To develop an innovative, explainable AI model for detecting anomalies in solar PV panels.
• To enhance the lifespan and power generation efficiency of PV systems through early fault detection.
• To provide a user-friendly tool for informed decision-making in PV system maintenance.

Main Methods:

• Utilized an enhanced Convolutional Neural Network (CNN) combined with the VGG16 architecture for anomaly detection.
• Implemented dataset balancing via oversampling and data augmentation to improve model robustness.
• Developed a user interface using PyQt5 for intuitive interaction and decision support.

Main Results:

• The AI model achieved high performance metrics: 91.46% accuracy, 98.29% specificity, and an F1 score of 91.67%.
• Successfully identified physical and electrical anomalies such as dust accumulation and bird droppings.
• The PyQt5 interface facilitated user-friendly operation and decision-making.

Conclusions:

• The developed explainable AI model effectively detects anomalies in solar PV panels, enhancing system efficiency.
• The approach contributes to prolonging the lifespan of photovoltaic systems and minimizing environmental risks.
• This AI-driven solution offers a promising method for proactive maintenance and management of solar energy infrastructure.