Fault Assessment and Early Performance Prediction of PV Module Using Machine Learning

An increase in demand for electrical energy in recent years have been observed due to the technological advancements. Satisfying the demand with fossil fuel-based energy generation has become a bottle neck situation. To address this issue, researches started to explore alternative source of energy. One such cost-effective, environment-friendly alternative energy source is solar photovoltaic (PV)-based power production. It is a worldwide accepted technology for extracting energy from renewable resources. Though the technology is harnessing with fossil fuel-based energy production, there are few challenges that need to be addressed in the photovoltaic (PV) technology. Now, industries have initiated an intensive study on cracks at the module stage in bulk solar panel cells. This is because, output of solar PV technology mainly depends on solar PV module. The damage, micro cracks, grain boundaries, and other solar cell flaws will have a direct impact on power generation and solar conversion effectiveness. In this work, an intensified review on various techniques employed for detecting micro cracks in polycrystalline solar cells has been done. In order to combine deep, semantically powerful features with low-level characteristics using a self-attention mechanism, a distinctive attribute of the fusion model is specifically proposed. A pretrained crack detector eases the accurate micro crack detection, where an approach to transfer learning based on MK-MMD to direct the instructional procedure for the fault detector has been employed. Such implementation enhances the efficacy of defect detection of a solar PV module.