Zhang, N.; Yang, G.; Hu, F.; Yu, H.; Fan, J.; Xu, S. A Novel Adversarial Deep Learning Method for Substation Defect Generation. Preprints2024, 2024051016. https://doi.org/10.20944/preprints202405.1016.v1
APA Style
Zhang, N., Yang, G., Hu, F., Yu, H., Fan, J., & Xu, S. (2024). A Novel Adversarial Deep Learning Method for Substation Defect Generation. Preprints. https://doi.org/10.20944/preprints202405.1016.v1
Chicago/Turabian Style
Zhang, N., Jingjing Fan and Siqing Xu. 2024 "A Novel Adversarial Deep Learning Method for Substation Defect Generation" Preprints. https://doi.org/10.20944/preprints202405.1016.v1
Abstract
The lack of defect image data is one of the main factors affecting the accuracy of supervised deep learning-based defect detection models. In response to the insufficient training data of defect images with complex backgrounds such as rust and surface oil leakage in substation equipment, leading to poor performance of the detection model, this paper proposed a novel adversarial deep learning model for substation defect generation: ADD-GAN. In comparison to existing generative adversarial networks, this model generates defect images based on effectively segmented local areas of substation equipment images, avoiding image distortion caused by global style changes. Additionally, the model utilizes a joint discriminator for overall image and defect image to address the issue of low attention to local defect areas, thereby improving the loss of image features. This enhances the overall quality of generated images as well as locally generated defect images, ultimately improving image realism. Experimental results demonstrate that the YOLOV7 object detection model trained on the dataset generated using the ADD-GAN method achieves an mAP of 81.5% on the test dataset, representing an improvement of 9.6% over the original dataset, 5.7% over traditional augmentation methods, and 7% over typical adversarial deep learning methods. This confirms that the ADD-GAN method can generate a high-fidelity dataset of substation equipment defects.
Keywords
Generation of defect images for substation equipment; GAN; Local region defect generation; Joint discriminator for overall image and defect image
Subject
Engineering, Energy and Fuel Technology
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
