An Appearance Abnormality Detection Method of Power Equipment Based on Image Comparison

Abstract

The invention discloses a method for detecting abnormal appearance of electric equipment based on image comparison. The steps are: normalizing the current inspection captured image, and then registering it with the historical inspection captured image at the same location and angle, and then Next, continue the regional segmentation of the two registered images, extract several features of each region image, fuse the several features, and finally calculate the difference between the corresponding features of the two images, and compare the difference with the set threshold. Determine whether the current inspection captured image is an abnormal image. The present invention has the advantage of realizing abnormal detection of various types of power equipment under the same algorithm framework. The algorithm compares two images taken by the inspection robot at the same stop at the same angle and at different times, and determines the current image with the same content. Regional changes in real-time images and historical inspection images can be used to detect abnormalities such as damage to power equipment and hanging foreign objects.

Description

technical field [0001] The invention relates to a method for detecting abnormal appearance of electric equipment based on image comparison. Background technique [0002] With the continuous expansion of the scale of my country's power network, the safe and reliable operation of substations / converter stations is very important, and the operating status of power equipment is one of the key factors that determine its safe and stable operation. Daily observation of equipment status in substation / converter station is one of the most arduous tasks in the inspection process. The main inspection contents of the on-duty personnel in the station include: suspension of foreign objects, damage and deformation of equipment, oil leakage of oil storage equipment, surface stains etc. [0003] With the development of computer science, image processing and recognition technology has achieved rapid development in both theoretical research and practical application. Based on the application of...

AI Technical Summary

Problems solved by technology

[0005] First, the current inspection robot equipped with image processing and recognition technology has limited detection functions, and most of the images taken by inspection still need to be observed and analyzed by operators at any time, which increases the burden on the on-site staff;

[0006] Second, some existing power equipment anomaly recognition algorithms need to establish different image template libraries for different equipment in substations / converter stations. The anomaly detection of different types of equipment requires the design of different image recognition algorithm frameworks, which consumes a lot of time and effort. manpower;

[0007] Third, using image processing and recognition technology to detect abnormalities in power equipment, it is first necessary to locate the position of the equipment to be inspected in the image. The accuracy of equipment positioning will directly affect the recognition effect of power equipment status detection, and positioning deviation will cause False positives and missed negatives occur, and there are certain security risks

Images