Analysis Of Rail Wear Detection Based On Image Processing

With the increasing speed of domestic railway traffic and the continuous expansion of railway mileage,the railway operation presents the phenomenon of high density and high rail load.Therefore,it is very important to strengthen the detection and Research on the track health parameters.It is necessary to carry out regular track wear detection and related deformation monitoring.However,for the track wear detection,the contact detection method is still adopted by domestic railway institutions most of the time.Although this method is simple and easy to popularize,it has long time to collect data,low efficiency and high labor cost.In view of this phenomenon,this paper adopts a track detection car developed by the project team and installs two 2D laser displacement sensors to collect the left and right tracks.This method is not only efficient,but also has high detection accuracy and low error caused by human.This paper first introduces the structure design of track detection trolley,then introduces the principle of 2D laser sensor and the application of image processing in this experiment.After calculating the track wear,the track is modeled in three-dimensional,and the deformation of track is analyzed.The specific research contents are as follows:The structure system,sensor system and data acquisition system of track detection car are introduced.The car body is made of steel structure,and the chassis is I-shaped.The sensors include 2D laser displacement sensor,inertial navigation system,laser radar and shaft angle encoder.The 2D laser displacement sensor adopts the optimization sensor produced by Elag company of Switzerland.When it works,it can emit a red light band to collect the data of track cross-section contour,and the image is clear Presented in the upper computer software.The image collected by 2D laser displacement sensor is processed simply.The complete image of track section is extracted through HSV image channel,and then filtered.As a non-linear filtering algorithm,median filtering can better retain the edge information of track section lines,which is the advantage of mean and Gaussian filtering.After removing the noise in the image,ROI region of interest is extracted.In this paper,the rail wear is mainly to detect the rail head area,so only the head image of the rail head area needs to be retained,which can greatly improve the calculation efficiency of the subsequent calculation.Finally,the skeleton of the processed image is extracted to improve the detection accuracy.After skeleton extraction,the surface of the skeleton is fitted,and the image of the rail head area is fitted by cubic uniform B-spline algorithm.The track image for wear is placed in the same coordinate system with the head image of the track head area to be detected,and the track wear data is calculated according to the proposed track wear algorithm.Compared with the traditional track wear detection method,this method can better improve the track wear status.Because the traditional detection method is two-point method,through the average value of two points to replace the wear value,it will produce a large error.At the same time,several fitting curves are used for 3D drawing with visual studio 2017 to get the relevant information of 3D surface of track.Aiming at the research of track deformation analysis,this paper proposes an analysis method of local deformation of 3D track curve based on MDP deformation detection algorithm.The specific experiments are carried out in the 100 meter track experimental line on campus.Through the experimental results,it is found that the wear of the track head section and the deformation detection in the three-dimensional image proposed in this paper can meet the experimental detection standards.