Lightweight and Improved U2 -Net Small Sample Recognition Algorithm for Coating Corrosion Law of Photovoltaic Power Generation Equipment in Salt Fields

In order to improve the recognition accuracy of coating corrosion in salt field photovoltaic power generation equipment, a lightweight improved U2 Net small sample recognition algorithm for the corrosion law of salt field photovoltaic power generation equipment coatings was studied. Analyze the corrosion patterns of coatings through long-term monitoring data. A lightweight improvement was made to the U2 Net network to address the unique characteristics of the salt field environment, and a CEA-U2-RFB 1Net network recognition model was established. Train recognition models using cross entropy loss functions and data augmentation techniques to identify the corrosion state of coatings. The experimental results show that this method outperforms the other three methods in terms of recognition accuracy, AUC-ROC curve performance, and recognition efficiency under the condition of increasing big data, and has good recognition performance and practical application value.