Application Of Infrared Image In Fault Detection Of Overhead Lines

In order to ease the contradiction between power supply and demand,China has vigorously promoted UHV technology in recent years.The UHV transmission has higher requirements on the reliability of the transmission process and does not allow power off maintenance.The automatic detection of high-voltage transmission lines based on infrared images can detect the equipment without power interruption and is more efficient and safe than traditional manual inspection.This is a good fit for the needs of power grid development.Based on the analysis of the basic principles of infrared imaging,the types of faults in overhead transmission lines are analyzed,including the experiments and analysis of temperature rise-time characteristics of 110 k V,220 k V zinc oxide surge arresters,220 kV coupling capacitors,and related electrical equipment.The temperature rise-time,time-to-phase temperature difference-time diagrams were plotted using the temperature measured by the infrared image and the statistical regularity of these devices during normal operation was analyzed.Focus on the image of the electrical equipment used in overhead lines for fault identification and hidden trouble investigation.In order to facilitate computer identification and improve the efficiency of fault identification,infrared images for these electrical devices require a series of processing.Including: graying,denoising,binarization,morphological processing,region segmentation,etc.Among them,denoising is a key part of image processing and is the basis for later fault diagnosis.There are many types of image noise,including salt and pepper noise,Gaussian noise,Rayleigh noise,uniform noise,and gamma noise.Among them,salt and pepper noise is a typical noise.There are many ways to filter out this kind of noise,such as mean filtering,median filtering,and Gaussian filtering.For these classic algorithms,an improved algorithm based on Gaussian filtering is proposed,and a brief principle analysis is performed on the improved method.The filter effect under median noise and multi-window median filter was compared by experiments.The peak signal-to-noise ratio was used as the detection index of the image filtering effect,which proved the superiority of this method.Then the binarization and morphological processing were performed on the filtered image,and the feasibility and practicability of the method were proved by the segmentation and feature recognition.