Research On Railway Track Information Extraction And Application Based On Point Cloud Data

The railway track is a key component in the railway infrastructure that needs to be surveyed and positioned,after the rail is accurately measured and positioned,the data presented in the form of a point cloud.Currently,vehicle-mounted LiDAR systems and track geometry measuring instrument are mainly used to collect and obtain railway track point cloud data.Among them,the point cloud data collected by the vehicle-mounted LiDAR system is large and contains a lot of invalid information,it needs to be processed to extract the information of the track position,and affected by the various error sources of the LiDAR system,the coordinate accuracy of the vehicle-mounted LiDAR point cloud data is low.The track geometry measuring instrument can directly obtain high-precision point cloud data of railway track,which can be directly applied to railway track deformation analysis.The acquisition,processing and analysis of railway track point cloud data are important means to ensure the safety of railway engineering operations.This thesis focuses on the extraction and application of railway track information based on point cloud data.The main contents include:Summarized the main error sources of the vehicle-mounted LiDAR system,the coordinate accuracy of the railway track point cloud measured by the vehicle-mounted LiDAR system and the track geometry measuring instrument is compared and analyzed,including the plane coordinate difference analysis and elevation difference analysis of the centerline point on the top of the left and right rails,the analysis results show that the difference between the coordinates of LiDAR point cloud and the coordinates of track point cloud measured by the track geometry measuring instrument is centimeter level.Aiming at the problem of unreliable seed point selection in the traditional point cloud filtering method based on surface fitting,some improvement measures are put forward in the process of seed point selection and grid height difference threshold calculation to improve the accuracy of the filtering method and its applicability in the railway scene point cloud filtering;On the basis of filtering,considering the unique geometric structure of the railway track in the railway scene,a track information extraction method considering the geometric characteristics of the track is studied.The experimental and error analysis results show that the filtering method can quickly filter out many invalid point clouds around the rail,and retain the rail point cloud more completely,the track information extraction method can accurately extract the left and right rail point cloud and track center line.Based on the high precision point cloud data of a domestic large-span cable-stayed bridge main channel railway track,this thesis analyzes the variation law of the track center horizontal deviation and vertical deviation under different measurement periods and different ambient temperatures.Aiming at the problem that the calculation process of track alignment and longitudinal level in traditional track static measuring method is complex and can not directly reflect the influence of track offset on track regularity,this thesis studies the feasibility and applicability of the new method of calculating the track alignment and longitudinal level by using the horizontal and vertical deviation of left and right track points in the analysis of track regularity of bridge railway,and the track quality index(TQI)is calculated according to the seven indexes of track gauge,cross level,twist of track,left and right track alignment and longitudinal level,the results are compared with the calculation results of TQI from the dynamic inspection data,it is proved that the data calculation and analysis method used in this thesis is feasible and reliable in ensuring the safety of long-span cable-stayed bridge railway engineering operations.