Research Of Contact Line Insulator Contamination Detection Technology Based On Machine Vision

In the operation process of electrified railway,the contact line insulator plays an irreplaceable role.It is an important guarantee for the safety operation of electrified railway.China has various topographic types and complicated climate.The insulator surface would easily accumulate filth and flashover occurs,which often leads to blackouts.For the safety operation of electrified railway,flashover accident is a great hidden trouble.Therefore,the detections of the polluted insulator surface and the insulator running state are very important.With the development of machine vision technology,the technology has become successful in many fields,such as face recognition,license plate recognition,vehicle tracking,textiles,packaging defect detection.With the advantages of non-contact,fast speed,high accuracy and high intelligence,machine vision is one of the researching hotspots currently.This dissertation mainly talks about how to apply machine vision to improve the efficiency and intelligent degree technology in the process of detection of contact line insulator.This dissertation mainly discusses four aspects of content: a.Identification and location of contact line insulator;b.Identification and classification of insulator contamination;c.Classification of insulator contamination grades;d.Type and grade record of insulator contamination.Firstly,the dissertation introduces how to recognize and locate the contact line insulator.In this dissertation,the image information of the insulator is sent to the computer by using the binocular camera.To reduce the interference of external factors,the image is grayed and the noise is removed.Then the SURF algorithm is used to extract the characteristics of insulators.Finally,the location information of insulators is obtained by using binocular ranging.These parts are the prerequisites for the insulator contamination detection.According to H,S,and V components of insulator images and feature extraction,insulator surface contamination is estimated.These features are used to realize the identification and classification of contamination,and the contaminated area is marked.The classification of the contamination level is to extract the characteristics of the collected contamination samples and to deal with them.Based on some prior knowledge,cluster analysis is used to cluster samples.Then,the contamination level of insulator is divided by the result of clustering.Finally,the database system is used to establish a set of recording system,which can be used to record the type,location information,contamination type and contamination level of the insulator.It can provide information for other studies.In this dissertation,binocular cameras,six degrees of freedom mechanical arm and other equipment would be used.In terms of implementation,Matlab would be used to verify the theory to ensure the correctness,and then use the software to achieve.The software is compiled by using Visual Studio 2013 and OpenCV.The results of laboratory tests show that the recognition and localization of the insulator is accurate with the support of the binocular camera.On this basis,using the methods of feature extraction and cluster analysis,it can realize the automation of classification of insulator contamination and contamination grade,and improve the efficiency of the insulator state detection.