Design And Research On Image Recognition And Fault Detection Of Aerial Insulator

With the country’s strategic deployment of smart power grid and power system,the construction of a highly information-based,automated,interactive and high-quality power network that can adapt to the digital era has been put on the agenda.China has a vast territory,a complex geographical environment,and long span and distance of transmission lines,resulting in low efficiency and heavy workload in daily power maintenance.Insulators are important components in power line inspection,and also the components with the highest failure rate.They exposed to the air for a long period of time are prone to failures such as self-explosion,cracks,and foreign objects,which seriously threaten the stable operation of the power system.In recent years,it has become a research hotspot to obtain the working status of important components in transmission lines by analyzing real-time images collected by UAV.In this thesis,the insulator images are collected by unmanned aerial vehicle(UAV)to extract the characteristics of the insulator string,design and research the insulator detection and positioning method of aerial image which realize the fault location detection of insulator.The main work and conclusions of this article are as follows:1.By analyzing the morphological characteristics of insulators and the mathematical characteristics in gray projection,a recognition method based on insulator morphological characteristics and gray projection is proposed.First,the insulator image is converted from RGB space to HSI space,and the S component is extracted for Gaussian filtering to obtain the image to be processed.Second,OTSU segmentation was performed on k-means clustering results.After morphological processing,the segmentation image of the insulator was obtained,and the connected domain was marked to obtain the rough insulator extraction image.Then,the thin length,roundness,and duty ratio are extracted from the morphological characteristics of the insulator,and the peak variance and the valley variance are extracted from the grayscale projection.and insulator search is carried out on coarse extraction area to realize accurate identification and location of insulator.Experiments show that this method can effectively locate and identify insulator strings in more complex environments.2.By analyzing the morphological characteristics of the faulty insulator,a fault detection method based on the proportion of insulator regions is proposed.Firstly,the location information of the fault insulator string is obtained by using the spatial information of the smallest external rectangle identified by the insulator,and the image to be processed is obtained after the tilt correction.Then,the insulator string is divided into an insulator fault area containing a single insulator piece with equal spacing,and the total number of pixels in each area and the number of insulator pixels are calculated by using the scan window.Finally,Searching the fault area of the insulator and marking the fault area by comparing the duty cycle of each insulator fault area,so as to accurately locate the fault area of the insulator string.Experiments show that the method can locate the fault position of insulator string accurately.