Infrared image segmentation and temperature extraction of catenary insulator based on Repvgg-Unet model

Infrared imaging can quickly detect abnormal temperature rise of contact network insulators for fault diagnosis. A Repvgg-Unet model-based insulator detection method for contact network temperature rise is proposed. The Repvgg-Unet model is constructed by combining Repvgg block with Unet and adding CBAM attention mechanism. The Repvgg-Unet model is used for image segmentation of railway contact network insulator infrared images and temperature reading through data processing for insulator heating fault detection. Finally, the performance of the model is validated by comparison of several tests. The Repvgg-Unet model was used to segment the contact network infrared insulator images, with improved accuracy and iou compared to the Unet model. The results show that the model has a better segmentation effect on insulators, which is important for improving the inspection of contact network insulators, reducing labour costs, improving equipment intelligence and ensuring safe operation of railway transmission lines.