Visual Detection For Fault Diagnoisis Of Power Equipment By UAV Line Patrol

With the rapid development of the high voltage power system,the technology of visual detection for fault diagnoisis of power equipment has become an important research direction.As the efficiency and security issues of traditional manual detection,the use of unmanned aerial vehicles(UAV)for line patrol has become a development trend.It is of great significance to realize the intelligent identification and detection by computer vision and image processing technology for the images of power equipment collected by UAV.The visual detection system is designed in this paper for the faults of the metal corrosion and the insulator breakage in power equipments.The main contents of this paper are as follows:(1)This paper studies the theory of the RGB color space model.Based on the unique color characteristics of the corrosion and insulator areas,the RGB component clustering center of the fault area is determined by sampling statistics.And through the similarity measure method,the effective distinction between the fault images and the non-fault images is realized.(2)This paper studies the methods of the gray scale and preprocessing algorithm for UAV color images.Based on the color characteristics of the fault areas,an improved super-green algorithm is proposed to achieve the gray scale of the images,while the color characteristics are effectively highlighted.The histogram equalization,median filter and image sharpening algorithm are used to preprocess the gray image and improve the quality of images.(3)This paper studies three image segmentation algorithms,including histogram threshold segmentation,iterative method segmentation and maximum interclass variance method.Based on the color component constraints of the two fault areas,an RGB component constraint segmentation method is proposed,and the image segmentation algorithm for the two kinds of faults is determined by lots of experiments.The morphological processing is used to achieve the error point elimination and hole filling in the binary images.(4)The judgment algorithm is designed for each type of fault on power equipments.For the metal corrosion,the fault judgment system is designed by dividing the segmented fault area by red and counting the area ratio.For the fault of insulator breakage,the bilateral contour difference matching algorithm is proposed to judge thestate of insulators based on the law of spatial arrangement of insulator rules.In this paper,the MATLAB and VS2010 platform is used to achieve a complete algorithm for fault detection of power equipments and tests experiments.The test sample database contains a total of 231 test images,including 105 pieces of corrosion images,87 insulator failure images and 39 fault-free images.The average detection rate of 92.1% is obtained in the detection of corrosion failure of gold,and the average detection rate of insulator damage is 91.2%.The validity of the algorithm is verified.