Research On Contamination Detection Technology Of Catenary Insulator Based On Embedded Vision

With the implementation of medium and long-term railway network planning in China,the modernization of railway industry has entered a period of rapid development.As an important component of the electrified railway power supply system,catenary insulators are prone to contamination by external pollution due to their long-term exposure to outdoor environments.Contaminated insulators can cause flashover under certain conditions,affecting the power supply and operational safety of railway system.At present,the contamination detection for catenary insulators mainly relies on manual inspection and image processing technology for automatic detection of pictures of insulators.Such a detection method is not only inefficient,but also has poor real-time performance,and cannot meet the current needs.Therefore,it is necessary to use machine vision technology combined with embedded hardware platform to conduct related research on automatic detection of the state of insulator.In this dissertation,a catenary insulator is taken as the research object,and a portable,low-cost and expandable detection system of contaminated insulator is designed.It mainly studies and analyzes image preprocessing,identification and contamination detection of insulator.Firstly,for the noise and uneven illumination caused by the camera to obtain images,the image denoising and enhancement algorithms are studied to improve the image acquisition quality and compare the effects.Secondly,owing to the characteristics of inflexible shape and fixed position of catenary insulator,the identification of insulator can be achieved using feature point matching.The point features of the insulator are extracted by Shi-Tomasi,SURF and FAST algorithms,and the effect and efficiency of above algorithms are compared.The matching effect and efficiency of the same feature points are analyzed and compared using different descriptors,and the insulator image is identified by distance constraint and similarity matching.On the basis of this,we extract and segment the contamination of insulator disk by using the characteristics that contamination will cause the discrepancy of insulator disk color,and quantitatively analyze the histogram difference of different contamination degrees.Finally,the embedded platform Raspberry Pi 3B + and USB camera are selected to build the visual detection platform.The transplantation and optimization of the insulator detection algorithm are completed,and the dynamic insulator recognition and pollution detection classification under the embedded platform are realized.In the implementation of the algorithm,Visual Studio 2013,OpenCV 2.4.9 and MATLAB are used to validate and analyze the algorithm on PC,and then the test of algorithmic migration is carried out on Raspberry Pi.The experimental results show that the system has good recognition effect and meets real-time requirements,which can provide a feasible reference for intelligent state detection and automatic cleaning of catenary insulators.