Fault Identification of MOA Based on Infrared Thermal Imaging

Metal oxide arrester (MOA), as an important device for limiting overvoltage in transmission lines, can ensure the safe and stable operation of the power system. The technology of using infrared instruments for troubleshooting arrester is relatively mature, but environmental factors can cause significant interference in infrared detection, and its accuracy is not high through visual observation. This article proposes an improved U-Net MOA segmentation algorithm to segment arresters in infrared images, and obtains temperature information from the images through operations such as dataset preprocessing, image segmentation, grayscale conversion, and temperature extraction. By combining the surface temperature method, the arrester in the infrared image can be diagnosed and the fault level and related treatment suggestions can be given.