Integrating Uav-Acquired Data Augmentation and Deep Learning for Solar Panel Defect Detection

Identifying defects in solar panels is crucial for improving the immediate efficiency in the installation, design, and manufacturing procedures. Conventional techniques entail the utilization of electron microscopes to obtain sample images, which are subsequently scrutinized by experts for any imperfections. Implementing this method is costly and requires a significant amount of time, especially when dealing with huge projects in megawatts and gegawatts. This work presents an automated approach for detecting faults in solar panels, utilizing sophisticated deep-learning methods to accurately identify and classify the defects. The EfficientNetB0 model was trained using aerial images obtained from Unmanned Aerial Vehicles (UAVs). Our methodology integrates advanced data augmentation techniques and the Robust Histogram Approximation (RHA) algorithm to minimize noise and improve the quality of training data. The EfficientNetB0 model attained a 92% accuracy rate, accompanied by elevated precision, recall, and F1 scores. This study emphasizes the enhancement of maintenance operations and the increase in the dependability and effectiveness of solar energy installations through our technique. Moreover, it highlights the capacity of sophisticated deep-learning methods in renewable energy and advocates for the Sustainable Development Goals (SDGs) established by the United Nations through the promotion of economically viable and environmentally sustainable energy alternatives.