A Survey of Solar Panel Surface Defect Detection Methods Based on Improved VGG-16 Model

Solar panels are the core components of photovoltaic power generation systems, and their quality is directly related to safety and power generation efficiency. Therefore, surface defect detection of solar panels is of great practical significance. In view of the inefficiency and high cost of manual detection, this paper proposes the use of convolutional neural networks (CNNs) for the automatic recognition and classification of solar panel images, which has been proven to be effective in image classification in the field of deep learning. Various optimization and adjustment strategies were used to update the classical CNN model, VGG-16. Improvements include the partial use of pre-trained models, the selection of appropriate optimization algorithms, pooling operations, the prevention of overfitting, and the implementation of early stopping strategies. By comparing the improved VGG-16 model with the classical VGG-16 model, it was found that the accuracy of the data increased from 97.65% to 99.13%. This shows that the improved VGG-16 model has optimal classification characteristics.