Machine learning approaches for automatic defect detection in photovoltaic systems

Abstract

Solar photovoltaic (PV) modules are prone to damage during manufacturing, installation, and operation which reduces their power conversion efficiency. This loss diminishes their positive environmental impact over the lifecycle. Continuous monitoring of PV modules during operation via images captured by unmanned aerial vehicles is essential to ensure prompt repair or replacement of defective panels to maintain high efficiencies. Coupled with computer vision techniques, this approach provides an automatic, non-destructive, and cost-effective tool for monitoring defects in PV plants. We review the current landscape of deep learning-based computer vision techniques used for detecting defects in solar modules. We compare and evaluate the existing deep learning approaches at different levels, namely the type of images, data collection and processing method, deep learning architectures employed, and model interpretability. Most approaches involve the use of convolutional neural networks with data augmentation or generative adversarial network-based techniques to enhance dataset size. We evaluate the deep learning approaches through techniques aimed at determining their interpretability, which reveals that the model focuses on the darker regions of the image to perform the classification. This exercise points out clear gaps in the existing approaches while laying the groundwork for mitigating these challenges when building new models. Finally, we conclude with the relevant research gaps that need to be addressed and approaches for progress in this field: integrating weather data and geometric deep learning with existing approaches for robustness and reliability; leveraging physics-based neural networks to build more domain-aware deep learning models; and incorporating interpretability for building trustworthy models.