A Survey on Artificial Intelligence based Photovoltaic Cell Defect Detection Techniques

With the increase in demand for solar energy in the form of photovoltaic (PV) panels have become an important component in sustainable energy applications. The visible defects (cracks, black spots, corrosion, etc.) in PV cells can occur during manufacturing times or while sustaining under external environmental conditions like rain, extreme temperature, humidity, etc. As a result, timely fault analysis in solar panels is essential in order to enhance both the lifespan and energy efficiency performance. The Electrolumince (EL) imaging technique is one of the popular technique used to scan PV modules in order to identify these critical defects like cracks. However, traditional PV fault detection methods rely on manual inspection and hence tedious, prone to human errors. This survey paper discusses the different methodologies and techniques using artificial intelligence (AI), Deep learning (DL), and Computer vision (CV) for quick, reliable, and efficient crack detection in PV panels using EL imaging analysis. Convolutional Neural Networks (CNNs), Residual Neural Networks (ResNet), and Visual Geometry Group (VGG) have demonstrated significant improvements in defect detection accuracy. This survey paper also discusses the standard datasets available in this domain and a comparative analysis of proposed techniques in the literature.