Research on Temperature Extraction of Porcelain Insulator Steel Cap and Zero Insulator Detection Based on Visible Light Image and Infrared Image

Abstract

Obtaining the temperature distribution of porcelain insulator steel cap is a key link in the patrol inspection of zero-value defects of transmission and distribution line insulators. It is difficult to obtain the temperature of insulator steel cap automatically in infrared image under complex background. Combining the respective advantages of visible and infrared images, insulator strings in visible light images are detected based on Faster-RCNN deep learning network algorithm, with a detection rate of 98.3%. The improved Sobel operator is used to detect the insulator edge based on the insulator strength edge feature. The rectangular shape feature and gradual spacing feature of insulator steel caps are used to extract and correct all steel caps. The coordinate conversion relationship between visible light and infrared image of insulator in the same scene is studied, and the steel cap extraction in infrared image is realized. When detecting zero-value defects of 110 kV and 220 kV insulators, this method can accurately extract the temperature of insulator steel cap and find zero-value insulators, which has good effectiveness and practicability.