Research On Multi-fault Detection Of Transmission Line Insulators Based On Visible And Infrared Aerial Images

In recent years,with the improvement of our people`s life quality,the electric power industry has played a key role in our society.Ensuring the power system`s stable operation has become a necessary requirement for people's life.And transmission line inspection is an important part of ensuring the power system`s stable working.With the development of UAV technology and computer vision technology,the traditional manual tower-climbing patrol mode has been gradually replaced by UAV inspection mode because of its time-consuming and laborious risk.This paper focuses on the insulators identification and fault detection in UAV inspection.Insulator image is acquired by mounting a dual-light camera on the UAV.Using computer vision theory technology to analyze the acquired image,extract the insulator area,and detect the insulator fault.The main contents of this paper are as follows:1.Preprocessing of aerial image of insulator.Firstly,gray-scale image data is processed to reduce the size of operation data.Then,according to the gray level rule of insulator data,the image is stretched piecewise and linearly to highlight the target insulator area.Finally,the image is filtered to de-noise the noise that may appear in the image data acquisition and transmission.2.Segmentation of aerial image of insulator.This paper propose an image classification method based on image texture complexity and gray distribution.For simple background images,a pixel sparse algorithm is used to reduce the size of operation data;for complex background images,image quality is optimized by transforming the image into HSI color space.And then the target features are highlighted by image filtering and image erosion.Aiming at the image quality problem after complex background processing,the Hough transform counting method is improved.A search algorithm in candidate regions is proposed to locate insulator strings.Finally,a hybrid algorithm of image open operation and maximum connected domain extraction is used in the central axis region to realize the segmentation of insulator region.3.Insulator fault detection is based on aerial image.Firstly,the orientation of the insulator axis in the image is corrected based on the slope of the insulator axis.Then the insulator string is divided into individual insulators by the distribution of the insulator transverse axis.For self-exploding faults,the irregularities are detected by using the monotony of the transverse distance between insulators;for insulator stain faults,an algorithm of horizontal comparison of insulator gray difference is used to detect the fault area;for insulator temperature abnormal faults,the temperature sensitivity of infrared image is converted to the feature of gray transformation,and the recognition algorithm is designed to detect the abnormalities.The above algorithms are verified by experiments in this paper,and the results are good.Finally,the results of insulator recognition and fault detection in aerial images are achieved.