Image Recognition And Tracking Of Railway Insulators

Due to the influence of the environment,the insulators of contact line will have potential danger of pollution flashover,which will affect the stability of the railway traction power supply system and equipment safety,and destroy the train operation order.In order to prevent of pollution flashover accidents,at present,the railway departments use KJ series insulators washing vehicles manipulated by workers to carry out the insulators hot washing operation,whose efficiency is better than the traditional manual cleaning,but it still cannot keep up with the rapid development of the electrified railway mileages and the current maintenance tasks.Automatic insulators washing equipment,which is still in market gaps and urgent state in the field of railway insulators maintenance,has become the research focus.The paper analyzes the washing conditions of washing insulators,puts forward two design ideas to realize the automatically washing and determines the automatically washing scheme with machine vision,and designs technological process of the automatically washing.The automatically washing system of insulator is designed,and five subsystems are listed.One of the most important,machine vision system,is designed in detail,and the acquisition,positioning and tracking functions are proposed to solve the positioning and control problem of automatically washing,and the paper designs the hardware platform of the machine vision system,whose key hardware are calculated and selected.This paper introduces the research content of machine vision,and analyzes the working conditions of the visual system of the washing vehicle,which is some features of the image and video of acquisition.Meanwhile,the paper designs the algorithm of machine vision module,which accomplish the all visual tasks in system through four main steps: image acquisition and preprocessing module,image processing and analysis,image servoing and series of images tracking.In this paper,after the mass center coordinates of insulators were obtained through image preprocessing,image segmentation and target positioning,a control method that is named image-based visual servoing system(IBVS)is designed,turning the image information into the control signal to adjust posture of the flushing cannon barrel and camera to aim insulators to complete the insulators washing.The main idea of this method is formulating characteristic functions and making difference between two characteristic functions minimal.The characteristic functions establish the relation between the image and the control quantity of the washing device through two Jacobi matrix and its pseudo inverse matrix,finally achieving control purposes.In this chapter,lastly,IBVS control method is verified in MATLAB,which control washing device and the accuracy meets the requirement.The control method can realize the automatic water washing and aiming at insulators by information form images.The insulators that were already positioned and aimed by vision system is tracked in video(in other words sequence images).If all the contents of the images are searched,a lot of redundant information will be generated,and the real-time performance is consequently affected.It is necessary to introduce a prediction mechanism of the change of the target location to reduce the computation.The idea of video tracking algorithm in this paper is using the features based detection and match,and combining with the Mean shift algorithm to track the insulator locations in the video.Lastly,the simulation experiment is designed to track the insulators by Mean shift be applied to algorithm for multi-group videos,and the results be analyzed show that the effectiveness of the algorithm is reasonable,which provides the algorithms for automatically washing insulators vehicle.