Research On Computing Railway Track Geometry Based On Point Cloud Data

The railway track is the basic equipment that directly carries the high-speed and stable operation of the train.Its smooth geometry is an important prerequisite to ensure the safe operation of the train.After the railway is put into operation,the railway track has been subjected to the repeated action of the wheel axle power and the impact of natural external forces such as shrinkage,expansion and settlement for a long time,resulting in the deformation of the track.Therefore,the smoothness of the geometric state of the track needs to be tested regularly.At present,3D laser scanning technology has become an important means of point cloud data acquisition,playing an increasingly important role,and has made some achievements in various application fields,but it is rarely used in the field of railway engineering.This paper attempts to study the calculation of track geometry parameters of track point cloud from the aspects of station layout,pre-processing of track point cloud,extraction of track related point cloud,calculation of track static geometry parameters.This paper mainly completes the following work:(1)There are two splicing methods for the point cloud measured by the ground 3D laser scanner: target splicing and control point splicing,using scanner in target splicing and control point splicing two different splicing modes,splicing experiment under different scanning distance and number of target balls,get the splicing precision under different splicing methods and configurations.By comparing the splicing accuracy and the splicing workload,etc,select a reasonable scanning and splicing scheme for the collection of railway point cloud,so as to maximize the scanning efficiency when the accuracy and density of the track point cloud meet the requirements.(2)There are some problems in the original collected orbit point cloud,such as noise,data redundancy,uneven point cloud,etc,point clouds need to be resampled.It is easy to make point cloud model deform when point cloud denoising,according to the railway track point cloud model,the noise point cloud is classified according to the geometric position,then,statistical filtering and Laplace filtering are used to denoise the isolated noise and the hybrid noise respectively.Then,statistical filtering and Laplace filtering are used to denoise the isolated noise and hybrid noise,respectively,to achieve the purpose of point cloud filtering and the railway track model is not deformed.Based on the point cloud after denoising,the main simplification algorithms,such as bounding box method,curvature sampling method,non-uniform grid clustering Gauss sphere mapping method,are used to simplify the railway track point cloud.Railway track point cloud has different properties for different methods.Among them,the uneven grid method shows high reduction rate and high operation efficiency,then homogenize the railway track point cloud,finally,resampling of point cloud is realized.(3)Aiming at the resampled 3D point cloud of railway track,in XYZ coordinate system,according to the geometry of the railway track point cloud model,the point cloud of the required track related position is extracted.These point clouds are usually planar,project the point cloud into different planes,Fit the linetype of the railway track point cloud,rotating the line,calculate the static track gauge,level,height and other smoothness parameters of the railway track every 0.65 m.The result can meet the required precision of railway maintenance value.