Steel rail geometric dimension detection method based on three-dimensional laser scanning

Abstract

The invention discloses a steel rail geometric dimension detection method based on three-dimensional laser scanning. The method comprises the following steps: S1, arranging a three-dimensional laser scanner and a prism ball; s2, establishing a design coordinate system, and registering prism ball point cloud coordinates collected by a three-dimensional laser scanner; s3, steel rail point cloud is collected, and the collected steel rail point cloud is preprocessed; s4, extracting section contour features of the steel rail; and S5, the size of the steel rail is detected and corrected. The three-dimensional laser scanning technology is adopted to detect the steel rail, compared with a traditional contact type measuring method, the measuring speed can be remarkably increased, the labor intensity of workers can be reduced, the continuous high-precision steel rail can be obtained through fitting by collecting a large amount of point cloud data, and the measured data accuracy is higher.

Description

Technical Field

[0001] The invention relates to the technical field of railway engineering, and in particular to a method for detecting geometric dimensions of rails based on three-dimensional laser scanning. Background Art

[0002] The geometric dimensions of rails have an important impact on the safety and comfort of trains, and are an important basis for high-speed train travel. The geometric quality of rails includes the geometric dimensions of rail cross-sections and the horizontality of track gauges. At present, the main method for detecting the horizontal spacing of rails and the horizontality between rails is to use a track ruler, which has a slow detection speed and the detection section is intermittent and discrete, which may lead to missed detection of dangerous sections to a certain extent. In addition, it has the characteristics of low efficiency and large workload. The cross-sectional parameters of rails will change to a certain extent due to wear during train operation or due to the collision of trains, which will affect the speed and safety of trains. Therefore, the detection of rail cross-sectional parameters is of great significance to the safe operation of trains.

[0003] At present, the main method for detecting rails is contact measurement, that is, using a track ruler to contact the rails for measurement. This method has low measurement accuracy, large manual labor, and low efficiency, which is difficult to meet the current requirements of rapid expansion and continuous improvement of railway scale. At present, three-dimensional laser scanning can quickly collect object point cloud data, and has the characteristics of wide scanning area, high accuracy and high efficiency. Therefore, in order to solve the problem of detecting the geometric dimensions of rails, a method for detecting the geometric dimensions of rails based on three-dimensional laser scanning is proposed. Summary of the invention

[0004] In order to overcome the deficiencies of the prior art, the purpose of the present invention is to provide a method for detecting rail geometry based on three-dimensional laser scanning to solve the problems of low measurement accuracy, large manual labor and low efficiency in current rail detection.

[0005] The technical solution adopted by the present invention is as follows:

[0006] A method for detecting rail geometric dimensions based on three-dimensional laser scanning comprises the following steps:

[0007] S1: Arrange a 3D laser scanner and a prism ball, and use the 3D laser scanner to collect the point cloud coordinates of the prism ball;

[0008] S2: Establish a design coordinate system, use the prism sphere coordinates in the design coordinate system as reference coordinates, align the prism sphere point cloud coordinates collected by the 3D laser scanner, and realize the conversion between the point cloud coordinates and the design coordinates;

[0009] S3: Collecting rail point clouds, preprocessing the collected rail point clouds, and obtaining the position of the rails in the design coordinate system;

[0010] S4: extract the technical characteristics of the rail and obtain the cross-sectional profile of the rail;

[0011] S5: Establish a BIM rail design model, measure the obtained rail cross-sectional profile dimensions, and use the designed rail cross-sectional dimensions as a reference to detect and correct the measured rail cross-sectional profile dimensions.

[0012] Compared with the prior art, the present invention has the following beneficial effects:

[0013] 1. The present invention adopts three-dimensional laser scanning technology. Through the station arrangement and prism ball arrangement, the point cloud data of the track is collected and the conversion between the point cloud coordinate system and the design coordinate system is realized. After the conversion, the point cloud is sliced ​​and the rail profile data is extracted to realize the detection of the geometric parameters of the rail. When the rail is detected in this way, the three-dimensional laser scanning can quickly collect a large amount of high point cloud data. Compared with the traditional contact measurement method, it can not only significantly improve the measurement speed and reduce the labor intensity of manual labor, but also can fit a continuous high-precision rail by collecting a large amount of point cloud data, so that the measured data is more accurate.

[0014] 2. The present invention establishes a BIM design model for rails and compares the measured actual parameters and coordinates with the design parameters. It can not only detect various geometric parameters of the rails, but also correct the geometric parameters of the rails according to the BIM design model. It can timely and accurately correct the deviation between the actual construction and the design, and ensure that the geometric dimensions of the rails strictly meet the design requirements.

[0015] As a preferred embodiment of the present invention, in step S1, the laser emitted by the three-dimensional laser scanner forms an incident angle α with the normal line of the target plane, α≤80°, and the calculation formula of the measurement point spacing D of the three-dimensional laser scanner is as follows:

[0016]

[0017] Where H is the horizontal distance between the scanner and the rail, and L is the height of the 3D laser scanner from the ground;

[0018] At least three prism balls are arranged, and the three prism balls are respectively located between the two rails and on the outer sides of the two rails.

[0019] Beneficial effects:

[0020] Since the distribution characteristics of the rails and the measurement accuracy of the 3D laser scanner are affected by the incident angle of the laser emitted by the 3D laser scanner, it is necessary to arrange the measuring stations of the 3D laser scanner to improve the accuracy of data collection. When the incident angle formed by the laser emitted by the 3D laser scanner and the normal of the target plane is greater than the limit incident angle, its measurement accuracy will decrease. This solution sets the limit incident angle to 80° and arranges the measuring stations of the scanner under the condition that it is less than or equal to the limit incident angle, which can ensure the high accuracy of the point cloud of the measured area, and thus ensure that the distribution characteristics of the collected rails are more accurate.

[0021] Since the prism ball is used to align the coordinates collected by the 3D laser scanner, in order to ensure the coordinate conversion accuracy of the entire scanning area, this solution arranges three prism balls between the two rails and on the outside of the two rails respectively. This can ensure that the positions of the prism balls are distributed at the edge and internal key positions of the scanning area, thereby ensuring that the data collected by the 3D laser scanner in the scanning area is more accurate and the obtained rail point cloud data is accurate.

[0022] As a preferred embodiment of the present invention, in step S2, the total station measures the coordinates of the three prism balls as (X1, Y1, Z1), (X2, Y2, Z2), and (X3, Y3, Z3), and the three-dimensional laser scanner collects the point cloud data of the three prism balls, and the least squares fitting method is used to extract the center coordinates of the prism balls (X1', Y1', Z1'), (X2', Y2', Z2'), and (X3', Y3', Z3'). Take (X1',Y1',Z1'),(X2',Y2',Z2'),(X3',Y3',Z3') as reference coordinates, construct the precise same-name points of (X1',Y1',Z1'),(X2',Y2',Z2'),(X3',Y3',Z3') and (X1,Y1,Z1),(X2,Y2,Z2),(X3,Y3,Z3), and transform (X1',Y1',Z1'),(X2',Y2',Z2'),(X3',Y3',Z3') into the design coordinate system based on the minimum distance between the corresponding points, so as to realize the conversion between point cloud coordinates and design coordinates.

[0023] As a preferred embodiment of the present invention, in step S3, the rail point cloud preprocessing includes using Gaussian filtering technology to reduce the noise of the point cloud, using a grid to segment the point cloud into block point clouds of a fixed size, and then segmenting the grid where the rail is located based on the point cloud density to obtain the rail position.

[0024] As a preferred embodiment of the present invention, the process of grid segmentation of point cloud includes projecting the collected rail point cloud to the XOY plane to form a two-dimensional point cloud, rotating the two-dimensional point cloud to make the rail point cloud parallel to the X axis, and the grid segmenting the XOY plane into block point clouds of size d×d, and then calculating the density ρ of the point cloud in each grid in the Y direction of each X, and the two grids with the largest density in the Y direction of each X are the grids where the two rails are located respectively, and finally the grids where the segmented rails are located are formed into a set to obtain the rail position on the XOY plane, and the calculation formula of the density ρ is:

[0025]

[0026] Where F represents the number of point clouds in a single grid, and M is the width of the rail section.

[0027] As a preferred embodiment of the present invention, in step S4, the process of extracting the technical features of the rail includes slicing the rail point cloud along its axis, projecting the sliced ​​micro-segment rail point cloud onto the YOZ plane to form a two-dimensional point cloud; extracting the rail cross-section edge point cloud on the YOZ plane, and then clustering and segmenting the extracted edge point cloud; fitting each edge point cloud using the least squares method to obtain an edge equation, jointly solving adjacent edge equations to obtain control points of the rail cross section, and reconstructing the rail cross-section profile according to the control points and the edge equations.

[0028] As a preferred embodiment of the present invention, the slice thickness is Δ, and the slice spacing is the width of the lower foot of the rail section; cluster segmentation uses the different curvatures of the straight line and the circular curve of the rail section as the criterion for edge segmentation. First, a point is randomly selected in the micro-segment rail projection point cloud, and b points around the point are selected for least squares fitting. When the fitting standard deviation is greater than 0.001, the point selection is re-performed until the fitting standard deviation meets the requirement, and the group of point clouds is stored separately, and then the step is repeated until all the point clouds of the cross section are grouped separately, thereby obtaining the point cloud of each edge line;

[0029] The calculation formula of slice thickness Δ is as follows:

[0030]

[0031] In the formula, b represents the number of points extracted, F i represents a randomly extracted point cloud, and R represents a point cloud with F i is the radius of the circle center, P i It represents the number of points in the query range, k represents the empirical parameter, and the value of k is between 3 and 5.

[0032] As a preferred embodiment of the present invention, in step S5, the detection of the rails includes:

[0033] Track gauge detection: The centroid coordinates (X, Y, Z) of the cross-section profile are obtained by the centroid calculation formula, where the X coordinate is the X value of the slice projection plane, the Y value is the mean of the corresponding control points of the cross-section profile, and the Z value is calculated by the centroid calculation formula. The centroid calculation formula is as follows:

[0034]

[0035] Where n represents the number of parts that the rail section is divided into, S represents the area of ​​each divided part, and z represents the distance between the centroid of each part and the lower edge of the rail.

[0036] The horizontal distance between the cross-sectional centroids of two rails of the same cross section is measured as Q, and the width of the upper end of the rail is P, and then the actual gauge can be calculated. The difference between the actual gauge and the designed gauge is compared with the specification to realize the detection of the rail gauge.

[0037] The calculation formula of track gauge l is as follows:

[0038] L=QP

[0039] The calculation formula of horizontal spacing Q is as follows:

[0040]

[0041] In the formula, (X L ,Y L ,Z L ) and (X R ,Y R ,Z R ) represent the centroid coordinates of the left rail and the right rail respectively;

[0042] Levelness detection: For the non-superelevation section of the rail, the elevation difference between the cross-sectional centroids of the two rails of the same cross section is compared with the specification to achieve the levelness detection of the non-superelevation section; for the superelevation section, the angle β between the cross-sectional centroids of the two rails of the same cross section is compared with the angle of the design model to achieve the levelness detection of the superelevation section; the calculation formula of the angle β between the cross-sectional centroids of the two rails of the same cross section is as follows:

[0043]

[0044] In the formula, Z L The Z axis coordinate of the centroid of the left rail section. R It represents the Z-axis coordinate of the centroid of the right rail section, and Q represents the horizontal distance between the centroids of the two rail sections.

[0045] As a preferred embodiment of the present invention, in step S5, the rail deviation correction includes:

[0046] Track gauge correction: The centroid coordinates of the section in the BIM design model are the centroid coordinates of the designed section, and the centroid coordinates of the rail in the point cloud model coordinate system are the measured centroid coordinates, which correspond to the centroid coordinates of the designed section at the rail position (X B ,Y B ,Z B ) as the starting point, and measure the centroid coordinates (X S ,Y S ,Z S ) is the end point, and the deviation correction amount T of each rail is obtained;

[0047] The calculation formula of the deviation correction amount T is:

[0048]

[0049] Level correction: The centroid coordinates (X B ,Y B ,Z B ) is the starting point, and the corresponding measured centroid coordinates (X S ,Y S ,Z S ) is the end point, and the elevation difference C between the designed cross-section centroid and the measured centroid is the required correction amount of the rail;

[0050] The calculation formula of elevation difference C is as follows:

[0051] C=Z B -Z S . BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is an overall flow chart of an embodiment of the present invention;

[0053] Figure 2 is a schematic plan view of the arrangement of measuring stations according to an embodiment of the present invention;

[0054] Figure 3 is a schematic elevation diagram of the arrangement of the measuring stations according to an embodiment of the present invention;

[0055] Figure 4 is a diagram of the arrangement of prism balls according to an embodiment of the present invention;

[0056] Figure 5 This is a schematic diagram of the rail point cloud segmentation principle of an embodiment of the present invention;

[0057] Figure 6 is a schematic diagram of a rail design coordinate system according to an embodiment of the present invention;

[0058] Figure 7 is a flow chart of rail profile reconstruction according to an embodiment of the present invention;

[0059] Figure 8Schematic diagram of the track gauge calculation principle of an embodiment of the present invention;

[0060] Fig. 9 is a schematic diagram of the track gauge correction principle of an embodiment of the present invention;

[0061] Fig.10 It is a schematic diagram of the horizontality correction principle of an embodiment of the present invention. DETAILED DESCRIPTION

[0062] Typical embodiments that embody the features and advantages of the present invention will be described in detail in the following description. It should be understood that the present invention can have various changes in different embodiments without departing from the scope of the present invention, and the descriptions and illustrations therein are essentially used for illustration purposes rather than for limiting the present invention.

[0063] This embodiment discloses a method for detecting the geometric dimensions of a rail based on three-dimensional laser scanning. Figure 1 As shown, the following steps are included:

[0064] S1: Arrange a 3D laser scanner and a prism ball, and use the 3D laser scanner to collect the point cloud coordinates of the prism ball.

[0065] 3D laser scanner layout: Since the distribution characteristics of the rails and the measurement accuracy of the 3D laser scanner are affected by the incident angle of the laser emitted by the 3D laser scanner, it is necessary to plan the layout of the 3D laser scanner station to improve the data collection accuracy. In this scheme, the layout of the 3D laser scanner station is as follows: Figure 2 and Figure 3 As shown, the measuring point spacing of the 3D laser scanner is D, the incident angle between the laser emitted by the 3D laser scanner and the normal of the target plane is α, α≤80°, and the calculation formula of the measuring point spacing D of the 3D laser scanner is as follows:

[0066]

[0067] Where H is the horizontal distance between the scanner and the rail, and L is the height of the 3D laser scanner from the ground.

[0068] Prism ball arrangement: The prism ball is used to register the coordinates collected by the 3D laser scanner so as to convert the coordinates later. To ensure the accuracy of coordinate conversion in the entire scanning area, this solution has three prism balls arranged. The three prism balls are arranged between the two rails and on the outside of the two rails, respectively. Figure 4As shown. The positions of the three prism balls are distributed at the edge and key positions of the scanning area, which can ensure that the data collected by the 3D laser scanner in the scanning area is more accurate. Among them, the 3D laser scanner collects the point cloud data of the three prism balls, and extracts the coordinates of the center of the Sanger prism ball through the least squares fitting method, which are (X1', Y1', Z1'), (X2', Y2', Z2'), (X3', Y3', Z3').

[0069] S2: Establish the design coordinate system and register the coordinates of the prism point cloud collected by the 3D laser scanner. For details, see Figure 6 As shown, the three prism ball coordinates (X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3) are measured in the design coordinate system by using the total station, and these coordinates are used as reference coordinates. The coordinates of the center of the prism ball (X1', Y1', Z1'), (X2', Y2', Z2'), (X3', Y3', Z3') collected by the 3D laser scanner are aligned according to the corresponding reference coordinates, that is, the precise same-name points of (X1', Y1', Z1'), (X2', Y2', Z2'), (X3', Y3', Z3') and (X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3) are constructed, and then converted to the design coordinate system with the minimum distance 0 between the corresponding points as the condition, so as to realize the conversion of point cloud coordinates and design coordinates.

[0070] S3: Use a 3D laser scanner to collect rail point clouds and pre-process the rail point clouds. First, project the collected rail point clouds onto the XOY plane to form a 2D point cloud. Use Gaussian filtering to reduce noise on the XOY plane point cloud. Then rotate the 2D point cloud to make it parallel to the X-axis. Then use a grid to segment the XOY plane into block point clouds of size d×d. Then segment the grid where the rails are located based on the point cloud density to obtain the rail positions. For details, see Figure 5 As shown, the density ρ is first calculated for each point cloud in each grid in the Y direction where each X is located. In the Y direction where each X is located, the two grids with the largest density are the grids where the two rails are located. These two grids are segmented, and finally the grids where the segmented rails are located are grouped together to obtain the rail positions on the XOY plane. The calculation formula for density ρ is:

[0071]

[0072] Where F represents the number of point clouds in a single grid, and M is the width of the rail section.

[0073] S4: Extract the profile features of the rail section. First, slice the rail point cloud along its axis with a slice thickness of Δ and a slice spacing of the rail section low foot width M. After the rail point cloud is sliced, the sliced ​​micro-segment rail point cloud is projected onto the YOZ plane, and then the edge point cloud of the rail section is extracted, and the extracted edge point cloud is clustered and segmented. Since the cross-section of the rail is mainly composed of two elements: straight lines and circular curves, where the curvature of the straight line is infinite and the curvature of the circular curve is a fixed value, the different curvatures of the straight line and circular curve of the rail section can be used as the standard for edge segmentation to achieve clustering segmentation of the rail section edge and obtain the point cloud of each edge.

[0074] The specific steps are as follows: first, randomly select a point in the micro-segment rail projection point cloud, and select b points around the point for least squares fitting. When the fitting standard deviation is greater than 0.001, reselect the point until the fitting standard deviation meets the requirements. The group of point clouds is stored separately, and then the step is repeated until all the point clouds of the cross section are grouped separately, thereby obtaining the point cloud of each edge line of the rail cross section. After obtaining the point cloud of each edge line, the least squares method is finally used to fit each edge line point cloud to obtain the edge line equation, and the adjacent edge line equations are jointly solved to obtain the control points of the rail cross section. The rail cross section is reconstructed based on the control points and edge line equations. For the specific rail contour reconstruction flow chart, see Figure 7 shown.

[0075] The calculation formula of slice thickness Δ is as follows:

[0076]

[0077] In the formula, b represents the number of points extracted, F i represents a randomly extracted point cloud, and R represents a point cloud with F i is the radius of the circle center, P i represents the number of points in the query range, and k represents the empirical parameter.

[0078] The larger the empirical parameter k value is, the more complete the features are, but the slice projection thickness Δ will also be larger, which will lead to redundant points in the projection section and reduce the reliability of the projected point cloud edge, affecting the processing rate and accuracy of the slice. Therefore, in this scheme, the empirical parameter k is set to 3 to 5, at which time the projection section generates fewer redundant points, and the slice projection can also better reflect the contour boundary.

[0079] S5: Establish a BIM rail design model, measure the obtained rail cross-sectional profile dimensions, and use the designed rail cross-sectional dimensions as a reference to detect and correct the measured rail cross-sectional profile dimensions.

[0080] Track gauge detection: Track gauge refers to the vertical distance between the inner sides of two rails. For the principle of track gauge measurement, see Figure 8 The centroid coordinates of the rail profile are (X, Y, Z), where the X coordinate is the X value of the slice projection plane, the Y value is the mean of the corresponding control points of the cross-section profile, and the Z value is calculated by the centroid calculation formula, which is as follows:

[0081]

[0082] Where n represents the rail section being divided into n parts, S represents the area of ​​each divided part, and z represents the distance between the centroid of each part and the lower edge of the rail.

[0083] According to the centroid coordinates, the centroid coordinates (X L ,Y L ,Z L ) and the centroid coordinates of the right rail (X R ,Y R ,Z R ), the horizontal distance Q between the cross-sectional centroids of two rails of the same cross-section and the width P of the upper end of the rail can be obtained, and then the actual track gauge l can be calculated. The calculation formula of the actual track gauge l is as follows:

[0084] L=QP

[0085] The calculation formula of horizontal spacing Q is as follows:

[0086]

[0087] By establishing a BIM rail design model and comparing the difference between the measured gauge and the designed gauge with the specification, the rail gauge can be detected.

[0088] Track gauge correction: The centroid coordinates of the section in the BIM design model are the designed section centroid coordinates, and the rail centroid coordinates in the point cloud model coordinate system are the measured centroid coordinates. For details, see Fig. 9 As shown, the centroid coordinates (X B ,Y B ,Z B ) as the starting point, and measure the centroid coordinates (X S ,Y S ,Z S ) is the end point, construct the vector corresponding to (X B -X S ,Y B -Y S ,Z B -Z S ) to obtain the horizontal displacement direction of the rail, and then calculate the deviation correction amount T of each rail;

[0089] The calculation formula of the deviation correction amount T is:

[0090]

[0091] Levelness detection: Rail levelness refers to the elevation difference between the left and right rails. For specific detection principles, see Fig.10 As shown. For the non-superelevation section of the rail, the elevation difference between the cross-sectional centroids of the two rails of the same cross section is compared with the specification to achieve the horizontal detection of the non-superelevation section; for the superelevation section, the angle β between the cross-sectional centroids of the two rails of the same cross section is compared with the design angle under the BIM design model to achieve the horizontal detection of the superelevation section; the calculation formula of the angle β between the cross-sectional centroids of the two rails of the same cross section is as follows:

[0092]

[0093] In the formula, Z L The Z axis coordinate of the centroid of the left rail section. R It represents the Z-axis coordinate of the centroid of the right rail section, and Q represents the horizontal distance between the centroids of the two rail sections.

[0094] Level deviation correction: Get the corresponding rail position under the BIM design model, and use the centroid coordinates (X B ,Y B ,Z B ) is the starting point, and the corresponding measured centroid coordinates (X S ,Y S ,Z S ) is the end point, and the calculated elevation difference C between the centroid of the design section and the measured centroid is the required correction amount of the rail. If the C value is positive, it means that the rail has moved downward, and if the C value is negative, it means that the rail has moved upward, thereby determining the displacement direction of the rail;

[0095] The calculation formula of elevation difference C is as follows:

[0096] C=Z B -Z S

[0097] The present invention collects point cloud data of the track and detects the rails through three-dimensional laser scanning technology. Compared with the traditional contact measurement method, it can not only significantly improve the measurement speed and reduce the labor intensity of manual work, but also fit continuous high-precision rails by collecting a large amount of point cloud data, making the measured data more accurate.

[0098] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A method for detecting geometric dimensions of rails based on three-dimensional laser scanning, characterized in that: The following steps are involved: S1: Arrange a 3D laser scanner and a prism ball, and use the 3D laser scanner to collect the point cloud coordinates of the prism ball; S2: Establish a design coordinate system, use the prism sphere coordinates in the design coordinate system as reference coordinates, align the prism sphere point cloud coordinates collected by the 3D laser scanner, and realize the conversion between the point cloud coordinates and the design coordinates; S3: Collecting rail point clouds, preprocessing the collected rail point clouds, and obtaining the position of the rails in the design coordinate system; S4: extract the technical characteristics of the rail and obtain the cross-sectional profile of the rail; S5: Establish a BIM rail design model, measure the obtained rail cross-sectional profile dimensions, and use the designed rail cross-sectional dimensions as a reference to detect and correct the measured rail cross-sectional profile dimensions.

2. The rail geometry detection method based on three-dimensional laser scanning according to claim 1, characterized in that: In step S1, the laser emitted by the three-dimensional laser scanner forms an incident angle α with the normal line of the target plane, α is less than or equal to 80°, and the calculation formula of the measurement point spacing D of the three-dimensional laser scanner is as follows: Where H is the horizontal distance between the scanner and the rail, and L is the height of the 3D laser scanner from the ground; At least three prism balls are arranged, and the three prism balls are respectively located between the two rails and on the outer sides of the two rails.

3. The rail geometry detection method based on three-dimensional laser scanning according to claim 1, characterized in that: In step S2, after the design coordinate system is established, the prism sphere coordinates (X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3) are measured in the design coordinate system by using a total station, and the point cloud data of the three prism spheres are collected by a 3D laser scanner. The sphere center coordinates (X1', Y1', Z1'), (X2', Y2', Z2'), (X3', Y3', Z3') of the prism sphere are extracted by using the least squares fitting method. Take (X1',Y1',Z1'),(X2',Y2',Z2'),(X3',Y3',Z3') as reference coordinates, construct the precise same-name points of (X1',Y1',Z1'),(X2',Y2',Z2'),(X3',Y3',Z3') and (X1,Y1,Z1),(X2,Y2,Z2),(X3,Y3,Z3), and transform (X1',Y1',Z1'),(X2',Y2',Z2'),(X3',Y3',Z3') into the design coordinate system based on the minimum distance between the corresponding points, so as to realize the conversion between point cloud coordinates and design coordinates.

4. The rail geometry detection method based on three-dimensional laser scanning according to claim 1, characterized in that: In step S3, the rail point cloud preprocessing includes using Gaussian filtering technology to reduce the noise of the point cloud, using a grid to segment the point cloud into block point clouds of a fixed size, and then segmenting the grid where the rail is located based on the point cloud density to obtain the rail position.

5. The method for detecting geometric dimensions of rails based on three-dimensional laser scanning according to claim 4, characterized in that: The process of grid segmentation of point cloud includes projecting the collected rail point cloud to the XOY plane to form a two-dimensional point cloud, rotating the two-dimensional point cloud to make the rail point cloud parallel to the X axis, and the grid divides the XOY plane into block point clouds of size d×d. Then, the density ρ is calculated for the point cloud in each grid in the Y direction of each X. The two grids with the largest density in the Y direction of each X are the grids where the two rails are located. Finally, the grids where the segmented rails are located are grouped together to obtain the rail position on the XOY plane. The calculation formula of density ρ is: Where F represents the number of point clouds in a single grid, and M is the width of the rail section.

6. The rail geometry detection method based on three-dimensional laser scanning according to claim 1, characterized in that: In step S4, the process of extracting the technical features of the rail includes slicing the rail point cloud along its axis, projecting the sliced ​​micro-segment rail point cloud onto the YOZ plane to form a two-dimensional point cloud; extracting the rail cross-section edge point cloud on the YOZ plane, and clustering and segmenting the extracted point cloud; fitting each edge point cloud using the least squares method to obtain the edge equation, jointly solving adjacent edge equations to obtain the control points of the rail cross section, and reconstructing the rail cross-section profile based on the control points and edge equations.

7. The rail geometry detection method based on three-dimensional laser scanning according to claim 6, characterized in that: The slice thickness is Δ, and the slice spacing is the width of the rail cross section. Cluster segmentation uses the different curvatures of the straight line and circular curve of the rail cross section as the criterion for edge segmentation. First, a point is randomly selected in the micro-segment rail projection point cloud, and b points around the point are selected for least squares fitting. When the fitting standard deviation is greater than 0.001, the point selection is re-performed until the fitting standard deviation meets the requirements. The group of point clouds is stored separately, and then the step is repeated until all the point clouds of the cross section are grouped separately, thereby obtaining the point cloud of each edge line. The calculation formula of slice thickness Δ is as follows: In the formula, b represents the number of points extracted, F i represents a randomly extracted point cloud, and R represents a point cloud with F i is the radius of the circle center, P i represents the number of points in the query range, ρ1 represents the point cloud density, k represents the empirical parameter, and the value of k is between 3 and 5.

8. The rail geometry detection method based on three-dimensional laser scanning according to claim 1, characterized in that: In step S5, the detection of the rails includes: Track gauge detection: The centroid coordinates (X, Y, Z) of the section are obtained by the centroid calculation formula, where the X coordinate is the X value of the slice projection plane, the Y value is the mean of the control points corresponding to the section profile, and the Z value is calculated by the centroid calculation formula. The centroid calculation formula is as follows: Where n represents the number of parts that the rail section is divided into, S represents the area of ​​each divided part, and z represents the distance between the centroid of each part and the lower edge of the rail. Measure the centroid coordinates of two rails of the same cross section respectively, and find the horizontal distance between the two centroid coordinates as Q, and the width of the upper end of the rail as P, and then calculate the actual track gauge. Compare the difference between the actual track gauge and the designed track gauge with the specification to realize the detection of the track gauge of the rail. The calculation formula of track gauge L is as follows: L=QP The calculation formula of horizontal spacing Q is as follows: In the formula, (X L ,Y L ,Z L ) and (X R ,Y R ,Z R ) represent the centroid coordinates of the left rail and the right rail respectively; Levelness detection: For the non-superelevation section of the rail, the elevation difference between the cross-sectional centroids of the two rails of the same cross section is compared with the specification to achieve the levelness detection of the non-superelevation section; for the superelevation section, the angle β between the cross-sectional centroids of the two rails of the same cross section is compared with the angle of the design model to achieve the levelness detection of the superelevation section; the calculation formula of the angle β between the cross-sectional centroids of the two rails of the same cross section is as follows: In the formula, Z L The Z axis coordinate of the centroid of the left rail section. R It represents the Z-axis coordinate of the centroid of the right rail section, and Q represents the horizontal distance between the centroids of the two rail sections.

9. The rail geometry detection method based on three-dimensional laser scanning according to claim 1, characterized in that: In step S5, the rail deviation correction includes: Track gauge correction: The centroid coordinates of the section in the BIM design model are the centroid coordinates of the designed section, and the centroid coordinates of the rail in the point cloud model coordinate system are the measured centroid coordinates, which correspond to the centroid coordinates of the designed section at the rail position (X B ,Y B ,Z B ) as the starting point, and measure the centroid coordinates (X S ,Y S ,Z S ) is the end point, and the deviation correction amount T of each rail is obtained; The calculation formula of the deviation correction amount T is: Level correction: The centroid coordinates (X B ,Y B ,Z B ) is the starting point, and the corresponding measured centroid coordinates (X S ,Y S ,Z S ) is the end point, and the elevation difference C between the designed cross-section centroid and the measured centroid is the required correction amount of the rail; The calculation formula of elevation difference C is as follows: WHAT B -WITH S 。

Images