Deep Learning-Based Photovoltaic (PV) Panels Fault Detection Using VGG16 Architecture

The study is aimed at the urgent need for accurate detection of Photovoltaic (PV) panel faults, using deep learning technology specifically exploiting the architecture of the VGG16 type. The primary task here is to create a strong model that will categorise photovoltaic panel fault images into these six different categories. To achieve this, a training set of 885 images is used and the VGG16 model is trained over 15 epochs with a batch size of 32 and its default learning for learning rate. Mainly through accuracy performance, the effectiveness of the proposed model is evaluated in accurately classifying PV-photovoltaic panel faults. The results show that the proposed VGG16 model achieves an impressive accuracy of 86% in PV-panel fault classification. This study suggests that the model can identify a wide range of fault categories, thereby demonstrating the potential usefulness of this technology in real-world applications for detecting and maintaining faults in photovoltaic panels. The study highlights the benefits of incorporating Deep Learning techniques, particularly the VGG16 architecture, into the IDS sub-project of the Solar Power System. This integration can improve efficiency, and reliability, and facilitate troubleshooting. Property management for renewable energy infrastructure is the natural next step for this research. Opportunities thus arise to improve system performance and guarantee that solar arrays produce optimal energy levels.