Detection and classification of photovoltaic cell cracks in solar energy systems using machine learning based approaches via

Electroluminescence (EL) imaging is the most widely used diagnostic technique to detect defects at all stages of solar panel manufacturing, installation, and operation. This method can potentially reduce power outages by detecting and repairing solar panel faults such as microcracks and finger line breaks. EL testing is a reliable inspection method, but EL images can be difficult to interpret due to complex fault patterns and heterogeneous backgrounds. As a result, evaluating defective cells and determining the severity of a problem requires specific knowledge, making these methods time-consuming to execute manually for each cell. Therefore, automatic visual inspection of solar cells becomes very important. In this study, a deep learningbased model is presented for automatic detection and classification of solar cell faults. CNN-based model is extracted to extract deep features. Deep features are evaluated on six different machine learning classifiers. The popular two-class ELPV dataset is used to test the proposed methodology. Keywords: Electroluminescence imaging, solar panels, deep learning, defect, classification