Virtual Track Construction Method For CRTSⅢ Track Slab Based On 3D Laser Point Cloud

The high-speed railway CRTS III slab ballastless track is a ballastless track independently developed by my country and has been widely used in China.After the CRTSIII track slab is laid,it is necessary to install standard fasteners and lay the rails,and then obtain the track measurement data.By repeatedly replacing the fastener adjustment parts,the track smoothness index can meet the opening and operation requirements.Due to factors such as track slab laying construction quality control,there is a large deviation between the actual spatial position of the rail bearing platform and the theoretical position,resulting in a large number of subsequent track fine-tuning,and the replacement rate of fasteners in some sections is as high as 100%.%.At present,laser scanning technology has the advantages of fast scanning speed and high precision,and can quickly obtain accurate point clouds of CRTSIII track plates.Therefore,a virtual track is constructed using the accurate CRTSIII track plate point cloud combined with the standard specification of the fastener system and rail parameters to provide basic data for track fine-tuning.For ballastless track,the fastener system is installed on the bearing platform,so the bearing platform is the key part to determine the position of the rail.Therefore,according to the point cloud of CRTS III track slab,which is mainly composed of track slab plane,track slab and noise points,a track slab extraction algorithm based on multidimension and sub-regional statistical filtering is proposed.The extraction of the bearing platform in the point cloud of the CRTS III track slab is completed.Firstly,the plane of the track plate is removed by plane filtering,and then a multi-size sub-regional statistical filtering algorithm is introduced to remove a large number of noise points between adjacent rail platforms in the line direction.Finally,the rail platforms are extracted according to the minimum bounding box.Because the algorithm needs to go through 3 steps,such as 2 filter processing and minimum bounding box extraction of rail platform,each step needs to traverse all points,which is inefficient.In order to further improve the extraction efficiency of the rail bearing platform,an algorithm for sequence extraction of bearing platform combined with track knowledge is proposed.The algorithm firstly extracts another rail bearing platform of the sleeper according to the acquired one rail bearing platform of a sleeper,and according to the positional relationship between the left and right rail bearing platforms of the same sleeper.Then search the center of the adjacent rail platform along the advancing direction,and extract the point clouds of the two rail platforms of the sleeper.Finally,the cycle continues until all the rail platforms are extracted.Since the algorithm for sequence extraction of bearing platform combined with track knowledge only needs to calculate the columnar neighborhood height difference of a few points along the line direction,its efficiency is much higher than that of the rail bearing platform extraction algorithm based on multisize subregional statistical filtering.Experiments were carried out on the point cloud data of the measured CRTS III track slab,and the efficiency of the algorithm for sequence extraction of bearing platform combined with track knowledge was 45 times that of the rail bearing platform extraction algorithm based on multi-size subregional statistical filtering.It can be used for fast extraction of rail bearing platform in largescale rail plate point cloud.The spatial position and attitude of the rail bearing platform determine the spatial position and attitude of the rail.Therefore,a robust algorithm for accurate determination of the spatial position and attitude of the rail bearing platform is proposed,which solves the problem of obtaining the precise spatial position and attitude of the rail bearing platform when there is a machining deviation.The algorithm uses the accurately registered rail bearing platform model,based on the positioning reference line of the outer jaw surface,to perform spatial alignment,so as to obtain the accurate spatial position and attitude of the rail bearing platform.Among them,a method for extracting the key surface of the rail bearing platform based on the maximum criterion is proposed,which reduces the adverse effect of the outlier points(OCTKP)connected to the key surface in the extraction of the key surface of the rail bearing platform,and the robustness of the extraction of the key surface of the rail bearing platform is improved.Experiment with the measured track plate point cloud.After the spatial position and attitude of the rail bearing platform are accurately determined,the absolute value of the mean value of the signed distance on the corresponding key surface is less than 0.25 mm,and the spatial position of the rail bearing platform can be accurately determined.The actual position of the rail is determined by the spatial position and attitude of the rail bearing platform,the specification of the fastener adjustment piece and the size of the rail.Therefore,on the basis of obtaining the precise spatial position and attitude of the rail bearing platform,a virtual rail construction method for rail fine-tuning is proposed,which solves the problem of virtual rail construction adapting to different types of fasteners and rail parameters.Based on the positional relationship between the rail platform,standard fasteners and rails,a local virtual rail model is constructed,and straight rails are used to connect adjacent rail platforms to obtain a virtual track for track fine-tuning.The virtual track is constructed from the point cloud of the measured track plate,and the gauge deviation is less than 0.8mm,and the lateral deviation and elevation deviation are less than 0.5mm,which can provide basic data for later track fine-tuning.