A subway shield tunnel ring joint identification method

Abstract

A subway shield tunnel ring joint identification method, comprising the steps of: according to the principle of cylindrical projection, converting the three-dimensional tunnel point cloud obtained by moving three-dimensional laser scanning into two-dimensional plane tunnel point cloud; according to the scanning line characteristics of the point cloud, the two-dimensional plane tunnel point cloud is divided into multiple sections; the average value of the reflection intensity value of all point clouds in each section is calculated, and the average value is saved in the data container M in ascending order according to the mileage value; all data in M are traversed, the difference between the average value of the reflection intensity value of each section and the adjacent section is calculated, and if the condition is met, it is saved in the new point cloud P; the point cloud P is clustered using the DBSCAN clustering algorithm, and the identification of the tunnel ring joint is realized according to the clustering result. The present application can quickly and accurately locate the position of the ring joint in the tunnel, realize the accurate identification and positioning of the ring joint position in the complex environment tunnel without bolt hole features, improve the accuracy and efficiency of the deformation detection of the subway tunnel, and improve the safety of the subway operation.

Description

Technical Field

[0001] This invention relates to a method for identifying circumferential joints in subway shield tunnels, belonging to the field of subway tunnel point cloud processing and application technology. Background Technology

[0002] Currently, subways have become a major means for many large cities to alleviate surface traffic pressure, improve traffic efficiency, and facilitate residents' lives. Subway tunnels are buried deep underground in the city, situated within a geological environment, and are constantly affected by natural factors such as geological movements, groundwater changes, seismic activity, and extreme weather. Construction work on the surface and the long-term movement of trains also affect the structural safety of subway tunnels. If the structural deformation of a subway tunnel exceeds a certain limit, it can cause serious safety hazards such as cracks in the tunnel walls, lining spalling, and track deformation, ultimately leading to safety accidents. Therefore, regular deformation detection and maintenance of subway tunnels are of great significance.

[0003] Compared to traditional single-point measuring instruments, lidar can quickly, efficiently, and accurately acquire three-dimensional point cloud data of the target. It is increasingly being applied to subway tunnel inspection tasks, as the acquired point cloud data accurately reflects the true three-dimensional condition within the tunnel. However, subway tunnels are often constructed using the shield tunneling method, resulting in a highly regular arrangement of tunnel segments. Subway tunnel deformation primarily stems from relative misalignment between these regularly arranged segments, leading to a series of structural defects. Therefore, accurately identifying and locating the annular joint position without sacrificing the accuracy of the subway tunnel point cloud data has become a critical issue that urgently needs to be addressed regarding tunnel deformation.

[0004] Existing methods for tunnel circumferential joint identification based on advanced algorithms and neural network models mainly include methods that identify joints based on the arrangement of bolt holes within the tunnel, methods that project the tunnel point cloud into a grayscale image and then use image grayscale transformation and line fitting to identify joints, and methods that use deep neural networks to identify regular features in the tunnel grayscale image. These methods directly or indirectly rely on bolt hole features in the tunnel point cloud and are relatively inefficient. When faced with complex tunnel environments without bolt holes, they struggle to accurately identify and locate joints, failing to meet the needs of overall structural inspection of subway tunnels. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method for identifying circumferential joints in subway shield tunnels. This method can identify and locate tunnel circumferential joints in complex tunnel environments without bolt holes, improving the accuracy and efficiency of subway tunnel deformation detection and ensuring safe subway operation.

[0006] To achieve the above objectives, the present invention provides a method for identifying annular joints in subway shield tunnels, comprising the following steps:

[0007] S1. Based on the principle of cylindrical projection, the three-dimensional tunnel point cloud obtained by moving three-dimensional laser scanning is converted into a two-dimensional planar tunnel point cloud;

[0008] S2. Based on the scan line characteristics of the tunnel point cloud obtained by mobile 3D laser scanning, the 2D planar tunnel point cloud is divided into N sections.

[0009] S3. Calculate the average value M of the point cloud reflection intensity values ​​for all points in each cross section. i (i = 1, 2, ..., N), and save the mileage values ​​in ascending order into the data container M;

[0010] S4. Traverse all data in M, calculate the difference between the average reflection intensity value of each section and the adjacent section, and if the condition is met, save the point cloud in the corresponding section to a new point cloud P.

[0011] S5. Use the DBSCAN clustering algorithm to cluster P, and identify tunnel circumferential joints based on the clustering results.

[0012] Furthermore, the process of converting the three-dimensional tunnel point cloud into a two-dimensional planar tunnel point cloud in step S1 is as follows:

[0013] S1.1 The scan line point cloud of the tunnel point cloud refers to all points with the same mileage value, i.e., y-coordinate. Ellipse fitting is performed on each scan line point cloud in the tunnel point cloud. Ellipse correction is then performed based on the parameters of the fitted ellipse, and the center of the scan line is moved to the origin. The specific process is as follows:

[0014] Let the general equation of an ellipse be: Ax 2 ++Bxy++Cy 2 ++Dx++Ey++F=0, A, B, C, D, E, F are unknown parameters, and constraint condition B 2 -4AC < 0 ensures the fitted curve is an ellipse; construct the design matrix:

[0015]

[0016] Construct the scatter matrix S = D T D; Construct the constraint matrix:

[0017]

[0018] Solving the generalized eigenvalue problem: Sp = λRp, where p = [ABCDEF] T λ is the eigenvalue to be solved, S is the scatter matrix; the coordinates of the center point. in Based on the center point coordinates, move the point cloud data center on the scan line to the origin, and modify the coordinates of any point to...

[0019] S1.2 Project the elliptical-corrected 3D tunnel point cloud data onto a 2D plane, and calculate the angle between the line connecting any point p on the scan line and the origin along the Z-axis.

[0020] S1.3 Calculate the corresponding arc length l = θ × r, where r is the tunnel design radius; modify the coordinates of p to... Project the 3D tunnel point cloud onto the Z=r plane.

[0021] Furthermore, the specific process of extracting the tunnel point cloud cross-section in step S2 is as follows:

[0022] S2.1. Based on the mileage coordinates (y-coordinates) of each point in the point cloud, divide and save point clouds with the same y-coordinate together to form a scanline point cloud.

[0023] S2.2 Calculate the average spacing s between all scan lines. The calculation formula is as follows: Where n represents the total number of scan lines. This represents the i-th scan line. This represents the (i-1)th scan line;

[0024] S2.3, Based on the circumferential joint width W in the tunnel design. s Calculate the cross-sectional width W d ,

[0025] S2.4, Based on the cross-sectional width W d The tunnel point cloud is divided into There are several cross-sections, and `round()` represents the rounding function. max This represents the maximum value of y in the tunnel point cloud. min This represents the minimum value of y in the tunnel point cloud.

[0026] Further, in step S3, the average value M of the point cloud reflection intensity values ​​for all points in each cross section is calculated. i The specific process for (i = 1, 2, ..., N) is as follows: traverse all points in the point cloud of each cross section and calculate the sum of the intensity values ​​of all points. Where j represents the total number of point clouds at the cross section, and the average intensity value of the point cloud at the cross section is... Save the mean point cloud intensity value at the i-th cross section as M. i .

[0027] Furthermore, the specific process of step S4 is as follows:

[0028] S4.1 Calculate section M i Difference from the average strength value of the previous section i = 2, 3, 4, ..., N-1, where M i M represents the mean intensity value of the point cloud at the i-th cross-section. i-1 This represents the average intensity value of the point cloud at the (i-1)th cross section;

[0029] S4.2, Calculate section M i Difference from the mean strength value of the next cross section i = 2, 3, 4, ..., N-1, where M i M represents the mean intensity value of the point cloud at the i-th cross-section. i++1 This represents the average intensity value of the point cloud at the (i+1)th cross section;

[0030] S4.3 Determine whether D1 and D2 satisfy the following conditions: If satisfied, it will be related to M. i The corresponding cross-sectional point cloud is saved to P, where t is a threshold that can be set freely.

[0031] Furthermore, the specific process of step S5 is as follows:

[0032] S5.1 Set the neighborhood radius Eps of the DBSCAN algorithm to 0.1~0.3, and the number of points within the neighborhood radius MinPts>=5;

[0033] S5.2. Use the DBSCAN algorithm to cluster P, and save each cluster in the clustering results separately. After clustering, P will have f clusters.

[0034] S5.3 Calculate the coordinates of the center point of each type of result. Where i = 1, 2, ..., T, T represents the point cloud contained in each result; the coordinate range of the i-th clustering result on the y-axis. i = 1, 2, 3, ..., f;

[0035] S5.4, if Δy i <0.3, calculate separately and In the formula, W r r represents the width of the lining ring in a tunnel design. i r represents the difference between the y-coordinate of the center point of the i-th type of result and the y-coordinate of the center point of the (i-1)-th type of result, and an integer multiple of the ring width. i ' represents the difference between the y-coordinate of the center point of the i-th result and the y-coordinate of the center point of the (i+1)-th result and an integer multiple of the ring width; round(x) represents rounding x to the nearest integer. <r i <0.1 and 0 <r i If ′<0.1, then p i (x i ,yi ,z i ) is denoted as the coordinate of the circumferential seam position;

[0036] S5.5, Based on the coordinates p of the circumferential seam position i (x i ,y i ,z i ) and W r Calculate the position coordinates p of the previous circumferential seam. i (x i ,y i -W r ,z i ) and the coordinates of the next circumferential seam position p i (x i ,y i ++W r ,z i This process is repeated to calculate the coordinates of all circumferential seams. Based on the coordinates of all circumferential seams, a point belonging to each circumferential seam is found, where the point in the circumferential seam satisfies... In the formula p y Let y represent the y-coordinate of any point p in the tunnel point cloud. i W represents the y-coordinate of the i-th circumferential seam position. s This is the width of the circumferential joint; this completes the identification of the tunnel's circumferential joints.

[0037] This invention converts three-dimensional tunnel point cloud data into two-dimensional planar data through cylindrical projection while preserving their correspondence. Then, a cross-section extraction module is used to divide the two-dimensional tunnel point cloud into sections. By calculating the average point cloud intensity value in the cross-section and capturing the difference in point cloud intensity values ​​at the circumferential joint, tunnel circumferential joint identification is achieved. This invention can quickly and accurately locate the position of circumferential joints in tunnels, even in complex environments without bolt hole features. This improves the accuracy and efficiency of deformation detection in subway tunnels, and enhances the safety of subway operations. Attached Figure Description

[0038] Figure 1 This is a flowchart of the method of the present invention;

[0039] Figure 2 This is a schematic diagram illustrating the transformation of a three-dimensional tunnel point cloud into a two-dimensional planar point cloud in this invention;

[0040] Figure 3 This is a schematic diagram of the results of cross-sectional extraction of two-dimensional planar point clouds in this invention;

[0041] Figure 4 This is a schematic diagram of the results after clustering the cross-section extraction results using the DBSCAN algorithm in this invention;

[0042] Figure 5This is a schematic diagram of the circumferential seam identification result after inverting the clustering results in this invention;

[0043] Figure 6 These are experimental results of the method of the present invention on point clouds of tunnels with bolt holes and point clouds without bolt holes, respectively. Detailed Implementation

[0044] The invention will now be further described with reference to the accompanying drawings.

[0045] like Figure 1 As shown, a method for identifying circumferential joints in subway shield tunnels includes the following steps:

[0046] S1. Based on the principle of cylindrical projection, the three-dimensional tunnel point cloud obtained by moving three-dimensional laser scanning is converted into a two-dimensional planar tunnel point cloud;

[0047] S2. Based on the scan line characteristics of the tunnel point cloud obtained by mobile 3D laser scanning, the 2D planar tunnel point cloud is divided into N sections.

[0048] S3. Calculate the average value M of the point cloud reflection intensity values ​​for all points in each cross section. i (i = 1, 2, ..., N), and save the mileage values ​​in ascending order into the data container M;

[0049] S4. Traverse all data in M, calculate the difference between the average strength value of each section and the adjacent section, and if the condition is met, save the point cloud in the corresponding section to a new point cloud P.

[0050] S5. Use the DBSCAN clustering algorithm to cluster P, and identify tunnel circumferential joints based on the clustering results.

[0051] Furthermore, the process of converting the three-dimensional tunnel point cloud into a two-dimensional planar tunnel point cloud in step S1 is as follows:

[0052] S1.1 The scan line point cloud in the tunnel point cloud refers to all points with the same mileage value, i.e., y-coordinate. Ellipse fitting is performed on each scan line point cloud in the tunnel point cloud. Ellipse correction is then performed based on the parameters of the fitted ellipse, and the scan line center is moved to the origin. The specific process is as follows:

[0053] Let the general equation of an ellipse be: Ax 2 +Bxy++Cy 2 +Dx+Ey+F=0, A, B, C, D, E, F are unknown parameters, and constraint B is given. 2 -4AC < 0 ensures the fitted curve is an ellipse; construct the design matrix:

[0054]

[0055] Construct the scatter matrix S = D T D; Construct the constraint matrix:

[0056]

[0057] Solving the generalized eigenvalue problem: Sp = λRp, where p = [ABCDEF] T λ is the eigenvalue to be solved, S is the scatter matrix; the coordinates of the center point. in Based on the center point coordinates, move the point cloud data center on the scan line to the origin, and modify the coordinates of any point to...

[0058] S1.2 Project the elliptical-corrected 3D tunnel point cloud data onto a 2D plane, and calculate the angle between the line connecting any point p on the scan line and the origin along the Z-axis.

[0059] S1.3 Calculate the corresponding arc length l = θ × r, where r is the tunnel design radius; modify the coordinates of p to... Projecting the 3D tunnel point cloud onto the Z=r plane, such as... Figure 2 As shown;

[0060] Furthermore, the specific process of extracting the tunnel point cloud cross-section in step S2 is as follows:

[0061] S2.1. Based on the mileage coordinates (y-coordinates) of each point in the point cloud, divide and save point clouds with the same y-coordinate together to form a scanline point cloud.

[0062] S2.2 Calculate the average spacing s between all scan line point clouds. The calculation formula is as follows: Where n represents the total number of scan lines. This represents the i-th scan line. This represents the (i-1)th scan line;

[0063] S2.3, Based on the circumferential joint width W in the tunnel design. s Calculate the cross-sectional width W d ,

[0064] S2.4, such as Figure 3 As shown, based on the cross-sectional width W d The tunnel point cloud is divided into There are several cross-sections, and `round()` represents the rounding function. max This represents the maximum value of y in the tunnel point cloud. min This represents the minimum value of y in the tunnel point cloud.

[0065] Further, in step S3, the average value M of the point cloud reflection intensity values ​​for all points in each cross section is calculated. i The specific process for (i = 1, 2, ..., N) is as follows: traverse all points in the point cloud of each cross section and calculate the sum of the intensity values ​​of all points. Where j represents the total number of point clouds at the cross section, and the average intensity value of the point cloud at the cross section is... Save the mean point cloud intensity value at the i-th cross section as M. i .

[0066] Furthermore, the specific process of step S4 is as follows:

[0067] S4.1 Calculate section M i Difference from the average strength value of the previous section i = 2, 3, 4, ..., N-1, where M i M represents the mean intensity value of the point cloud at the i-th cross-section. i-1 This represents the average intensity value of the point cloud at the (i-1)th cross section.

[0068] S4.2, Calculate section M i Difference from the mean strength value of the next cross section i = 2, 3, 4, ..., N-1, where M i M represents the mean intensity value of the point cloud at the i-th cross-section. i++1 This represents the average intensity value of the point cloud at the (i+1)th cross section;

[0069] S4.3 Determine whether D1 and D2 satisfy the following conditions: If satisfied, it will be related to M. i The corresponding cross-sectional point cloud is saved to P, where t is a freely set threshold. Further, the specific process of step S5 is as follows:

[0070] S5.1 Set the neighborhood radius Eps of the DBSCAN algorithm to 0.1~0.3, and the number of points within the neighborhood radius MinPts>5;

[0071] S5.2, such as Figure 4 As shown, the DBSCAN algorithm is used to cluster P, and each cluster in the clustering results is saved separately. After clustering, P yields f clusters.

[0072] S5.3 Calculate the coordinates of the center point of each type of result. Where i = 1, 2, ..., T, T represents the total number of points contained in each cluster, and the coordinate range of the i-th cluster result on the y-axis is... i = 1, 2, 3, ..., f;

[0073] S5.4, if Δy i<0.3, calculate separately and

[0074] In the formula, W r r represents the width of the lining ring in a tunnel design. i r represents the difference between the y-coordinate of the center point of the i-th type of result and the y-coordinate of the center point of the (i-1)-th type of result, and an integer multiple of the ring width. i ' represents the difference between the y-coordinate of the center point of the i-th result and the y-coordinate of the center point of the (i+1)-th result and an integer multiple of the ring width; round(x) represents rounding x to the nearest integer. <r i <0.1 and 0 <r i If ′<0.1, then p i (x i ,y i ,z i ) is denoted as the coordinate of the circumferential seam position;

[0075] S5.5, such as Figure 5 As shown, based on the coordinates p of the circumferential seam position... i (x i ,y i ,z i ) and W r Calculate the position coordinates p of the previous circumferential seam. i (x i ,y i -W r ,z i Coordinates p of the position of the ring and the next ring seam i (x i ,y i +W r ,z i This process is repeated to calculate the coordinates of all circumferential seams. Based on the coordinates of all circumferential seams, a point belonging to each circumferential seam is found. A point within a circumferential seam should satisfy the following condition: In the formula, p y Let y represent the y-coordinate of any point p in the tunnel point cloud. i W represents the y-coordinate of the i-th circumferential seam position. s It is the width of the circumferential joint, which completes the identification of the circumferential joint in the tunnel.

[0076] like Figure 6 As shown, to make the results more intuitive, the tunnel point cloud was converted into a grayscale image, and the ring seam identification results were marked with red lines. Figure 6 (a) is a diagram showing the results of circumferential joint identification on the point cloud of a tunnel with bolt hole features according to the present invention. Figure 6(b) is a diagram showing the results of ring joint identification on bolt hole feature point cloud according to the present invention. It can be seen that the present invention does not rely on bolt hole features in tunnel point cloud. When facing complex and diverse tunnel environments, it can achieve tunnel ring joint identification and positioning with high efficiency and high accuracy, providing more reliable technical support for subsequent tunnel structure analysis and safety monitoring.

Claims

1. A method for identifying circumferential joints in subway shield tunnels, characterized in that, Includes the following steps: S1. Based on the principle of cylindrical projection, the three-dimensional tunnel point cloud obtained by moving three-dimensional laser scanning is converted into a two-dimensional planar tunnel point cloud; S2. Based on the scan line characteristics of the tunnel point cloud obtained by mobile 3D laser scanning, the 2D planar tunnel point cloud is divided into... One cross-section; S3. Calculate the average value of the point cloud reflection intensity for all points in each cross section. The mileage values ​​are saved to the data container in ascending order. middle; S4. Traverse all data in M, calculate the difference between the average reflection intensity value of each section and the adjacent section, and if the condition is met, save the point cloud in the corresponding section to a new point cloud P. S5. Use the DBSCAN clustering algorithm to... Clustering is performed, and the tunnel annular joints are identified based on the clustering results; The specific process of step S4 is as follows: S4.1 Calculation of cross-section Difference from the average strength value of the previous section ,in, Let represent the mean reflection intensity value of the point cloud at the i-th cross-section. This represents the reflection intensity value of the point cloud at the (i-1)th cross section; S4.2 Calculation of cross-section Difference from the mean value of reflection intensity of the next cross section ,in, Let represent the mean intensity value of the point cloud at the i-th cross-section. This represents the reflection intensity value of the point cloud at the (i+1)th cross section; S4.3, Judgment and Does it meet the requirements? If satisfied, it will be with The corresponding cross-sectional point cloud is saved to P, where t is the set threshold. The specific process of step S5 is as follows: S5.1 Set the neighborhood radius of the DBSCAN algorithm to Eps = 0.1~0.3, and the number of points within the neighborhood radius to MinPts > 5; S5.

2. Use the DBSCAN algorithm to cluster P, and save each cluster in the clustering results separately. After clustering, P will have f clusters. S5.3 Calculate the coordinates of the center point of each type of result. ,in T represents the point cloud contained in each cluster result; the coordinate range of the i-th cluster result on the y-axis. ; S5.4, if Calculate separately and In the formula, This indicates the width of the lining ring in the tunnel design. This represents the difference between the y-coordinate of the center point of the i-th type of result and the y-coordinate of the center point of the (i-1)-th type of result, and an integer multiple of the ring width. This represents the difference between the y-coordinate of the center point of the i-th result and the y-coordinate of the center point of the (i+1)-th result and an integer multiple of the ring width. This indicates rounding x to the nearest whole number; if 0 < And 0 < Then Recorded as the coordinates of the circumferential seam position; S5.5, Based on the coordinates of the circumferential seam position and the design width of the tunnel lining ring Calculate the position coordinates of the previous circumferential seam and the coordinates of the next circumferential seam position Calculate the coordinates of all circumferential seams in this manner; based on the coordinates of all circumferential seams, find the point belonging to each circumferential seam, where the point in the circumferential seam satisfies... In the formula, Represents any point in the tunnel point cloud y-coordinate, This represents the y-coordinate of the i-th circumferential seam position. It is the width of the circumferential joint, which completes the identification of the circumferential joint in the tunnel.

2. The method for identifying circumferential joints in subway shield tunnels according to claim 1, characterized in that, The process of converting the three-dimensional tunnel point cloud into a two-dimensional planar tunnel point cloud in step S1 is as follows: S1.1 The scan line point cloud of the tunnel point cloud refers to all points with the same mileage value, i.e., y-coordinate. Ellipse fitting is performed on each scan line point cloud in the tunnel point cloud. Ellipse correction is then performed based on the parameters of the fitted ellipse, and the center of the scan line is moved to the origin. The specific process is as follows: Let the general equation of an ellipse be: A, B, C, D, E, and F are unknown parameters and constraints. Ensure the fitted curve is an ellipse; construct the design matrix: ； Constructing a scatter matrix Construct the constraint matrix: ； Solving the generalized eigenvalue problem ,in , These are the eigenvalues ​​to be solved, S is the scatter matrix, and the coordinates of the center point are... ,in , Based on the center point coordinates, move the point cloud data center on the scan line to the origin, and modify the coordinates of any point to... ; S1.2 Project the elliptical-corrected 3D tunnel point cloud data onto a 2D plane and calculate the value of any point on the scan line. The angle between the line connecting the origin and the Z-axis ; S1.3 Calculate the corresponding arc length , Design radius for the tunnel; The coordinates are modified to Projecting the 3D tunnel point cloud onto On a plane.

3. The method for identifying circumferential joints in subway shield tunnels according to claim 1, characterized in that, The specific process of step S2 is as follows: S2.

1. Based on the mileage coordinates (y-coordinates) of each point in the point cloud, divide and save point clouds with the same y-coordinate together to form a scanline point cloud. S2.2 Calculate the average spacing between all scan line point clouds. The calculation formula is: Where n represents the total number of scan lines, This represents the i-th scan line. This represents the (i-1)th scan line; S2.3, Based on the circumferential joint width in the tunnel design. Calculate the cross-sectional width , ; S2.4, Based on the cross-sectional width The tunnel point cloud is divided into One cross-section, This represents the rounding function. This represents the maximum value of the y-coordinate in the tunnel point cloud. This represents the minimum y-coordinate in the tunnel point cloud.

4. The method for identifying circumferential joints in subway shield tunnels according to claim 1, characterized in that, In step S3, the average value of the point cloud reflection intensity values ​​for all cross sections is calculated. The specific process is as follows: traverse all points in the point cloud of each cross section and calculate the sum of the intensity values ​​of all points. ,in This represents the total number of point clouds at the cross-section, and the average point cloud intensity value at the cross-section. Save the average point cloud intensity value at the i-th cross-section as... .

Images