Defect image generation method, device and equipment for photovoltaic module EL detection

By preprocessing and generating model training on defective images of photovoltaic modules, expanding and fusing image data, the problem of insufficient data for existing photovoltaic modules is solved, and the accuracy and reliability of detection are improved.

CN119723242APending Publication Date: 2025-03-28SEARI ELECTRIC TECH CO LTD
0 Cites 1 Cited by

Patent Information

Application Number
CN202411792499.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-05
Publication Date
2025-03-28

AI Technical Summary

Technical Problem

Due to insufficient sample data in the existing photovoltaic module EL detection model, the effect of identifying foreign object defects is inaccurate, and the missed detection phenomenon occurs frequently, reducing the overall reliability of photovoltaic module EL detection.

Method used

By preprocessing the first defect image of the photovoltaic component with foreign object defects, training samples for training an initial defect image generation model based on the target generation adversarial network are generated. Then, the initial defect image generation model is trained based on the training samples and category labels, and a second defect image covering a variety of defect types is generated. Subsequently, the number of second defect images is expanded to obtain a plurality of expanded third defect images, and is fused with the normal image to generate a plurality of target defect images.

Benefits of technology

By generating more diverse defect images, the generalization performance of the EL detection model is significantly improved, so that it can better adapt to the EL detection needs of complex photovoltaic modules, and improve the accuracy and reliability of EL detection of photovoltaic modules.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a defect image generation method and device for photovoltaic module EL detection, computer equipment and a storage medium, and the method comprises the steps: carrying out the preprocessing of a first defect image of a photovoltaic module with a foreign matter defect, and obtaining a training sample for training an initial defect image generation model based on a target generative adversarial network; training an initial defect image generation model according to a training sample and a category label of the training sample to obtain a trained defect image generation model, generating second defect images through the defect image generation model, and expanding the number of the second defect images to obtain multiple expanded third defect images; and finally performing fusion processing on the plurality of expanded third defect images and a normal image of the photovoltaic module without foreign matter defects to generate a plurality of target defect images. According to the invention, the defect image for EL detection of the photovoltaic module can be accurately generated, and the accuracy and reliability of EL detection of the photovoltaic module are improved.
Need to check novelty before this filing date? Find Prior Art