Publicação
Artificial Intelligence for Fault Detection in Photovoltaic Panels
| Resumo: | This paper presents an Artificial Intelligence solution for fault detection and classification in photovoltaic systems. The proposed tool integrates electrical and visual analysis methods, including I-V curve analysis, direct difference measurement, infrared thermography, electroluminescence imaging, and visual inspection. These methods are enhanced by deep learning models, which achieve high accuracy in identifying and diagnosing faults. A Python-based web application provides users with an intuitive interface for real-time data processing and fault classification. Experimental results demonstrate the tool’s effectiveness, with neural network models achieving accuracy levels exceeding 98% in electrical methods and over 90% in visual methods. By optimizing fault detection processes, the tool reduces maintenance costs, minimizes downtime, and enhances the operational reliability of photovoltaic systems. This research represents a significant step toward scalable, automated maintenance solutions, ensuring photovoltaic systems’ sustainability and efficiency in the transition to renewable energy. |
|---|---|
| Autores principais: | José, D.F. |
| Outros Autores: | Janeiro, Fernando M.; Pires, V.F.; Pires, A.J.; Martins, J.F. |
| Assunto: | Fault detection Photovoltaic systems Artificial intelligence Deep learning Renewable energy |
| Ano: | 2025 |
| País: | Portugal |
| Tipo de documento: | palestra |
| Tipo de acesso: | acesso restrito |
| Instituição associada: | Universidade de Évora |
| Idioma: | inglês |
| Origem: | Repositório Científico da Universidade de Évora |
| Resumo: | This paper presents an Artificial Intelligence solution for fault detection and classification in photovoltaic systems. The proposed tool integrates electrical and visual analysis methods, including I-V curve analysis, direct difference measurement, infrared thermography, electroluminescence imaging, and visual inspection. These methods are enhanced by deep learning models, which achieve high accuracy in identifying and diagnosing faults. A Python-based web application provides users with an intuitive interface for real-time data processing and fault classification. Experimental results demonstrate the tool’s effectiveness, with neural network models achieving accuracy levels exceeding 98% in electrical methods and over 90% in visual methods. By optimizing fault detection processes, the tool reduces maintenance costs, minimizes downtime, and enhances the operational reliability of photovoltaic systems. This research represents a significant step toward scalable, automated maintenance solutions, ensuring photovoltaic systems’ sustainability and efficiency in the transition to renewable energy. |
|---|
Atividades financiadas
Não foram encontrados projetos financiados.