Crack Detection and Output Power Analysis in Solar Panels Using DANet Algorithm

Solar energy, derived from the sun’s radiation, is the best renewable resource powering various applications globally. Solar panels harness sunlight, generating clean electricity for residential and industrial sectors. Recent studies have indicated that cracks covering just 2% of a panel’s surface area can result in an efficiency decrease of around 6%. Moreover, severe cracks or damage can lead to even more significant reductions in output, potentially up to 20% or more. So efficient maintenance and timely detection of cracks in solar panels is critical for effective output. Manual crack detection in solar panels is hindered by the complexity of crack patterns, variations in lighting, and the potential for human error. It is time-consuming, lacks consistency, and may miss small or obscured cracks, impeding effective maintenance. So for effective crack detection deep learning model is used. Initially the images (obtained using EL imaging technique) is given as input to the model and then based on the crack segmentation power analysis is done. In this paper, DANet algorithm is used for segmentation of cracks in solar panel. Dual Attention Network enhances feature representation in deep learning by considering both spatial and channel-wise dependencies for improved performance. After the process of segmentation, it’s output is given for power analysis to know the severity of cracks.