High-speed rail tunnel no-survey mark section deformation resolving method

Abstract

The invention provides a measurement mark-free section deformation resolving method for a high-speed rail tunnel, and relates to the technical field of tunnel detection. The method comprises the following steps: acquiring point cloud data of a high-speed rail tunnel, and carrying out mileage alignment on the point cloud data and datum point cloud data; performing section registration on the point cloud data and the reference point cloud data; establishing a corresponding relation of two periods of point cloud data through a nearest neighbor point of any point position in the datum point cloud data, and setting a displacement difference threshold value and an intensity grading threshold value of a point pair based on the corresponding relation; performing clustering analysis on the point clouds exceeding the displacement difference threshold value or the intensity grading threshold value based on a target clustering algorithm, and generating a potential deformation area continuous mask; the potential deformation area continuous mask represents the area where the high-speed rail tunnel has diseases. According to the method, the detection efficiency and the detection accuracy of high-speed rail tunnel deformation detection can be improved.

Description

Technical Field

[0001] The present application relates to the field of tunnel detection technology, and in particular to a method for calculating deformation of a high-speed railway tunnel section without measurement marks. Background Art

[0002] With the rapid expansion of my country's high-speed rail network, tunnels, as critical infrastructure for high-speed trains, are facing increasing scrutiny. Their structural health is directly linked to operational safety and service life. Currently, high-speed rail tunnel defect detection relies primarily on traditional methods such as manual inspections and total station cross-sectional measurements. These methods have exposed systemic flaws in engineering practice. Manual inspections require personnel to enter the closed tunnel environment and visually inspect the lining surface with simple tools. This is not only inefficient and has limited coverage, but also results in inconsistent interpretation of defects due to subjective experience. Hidden defects such as fine cracks and initial spalling are easily missed, while the severity of defects such as lining seepage and hollowing is difficult to quantify and record. This lacks a reliable data basis for multi-period deformation trend analysis.

[0003] While cross-sectional measurements of high-speed rail tunnels using total stations can obtain quantitative data, they require the installation of instruments at each section within the tunnel. A single operation can only capture sparse, discrete measurement points (typically 30-50 points), making it impossible to fully characterize the continuous deformation characteristics of the tunnel's inner wall. Furthermore, measurement is time-consuming, and full-coverage inspections of long tunnels, stretching tens of kilometers, require interrupting traffic for several days, making it difficult to meet the needs of high-frequency, real-time monitoring. Consequently, existing technologies for deformation detection in high-speed rail tunnels suffer from low efficiency and accuracy. Summary of the Invention

[0004] Based on this, the present application provides a method for solving the deformation of high-speed railway tunnel sections without measurement marks. Based on the multi-period data comparison of mobile laser scanning technology, the measurement data of the high-speed railway tunnel is fused in multiple dimensions to improve the detection efficiency and accuracy of deformation detection, thereby realizing the tunnel structure status assessment and providing full-chain technical support for the safe operation and maintenance of the tunnel structure.

[0005] In a first aspect, the present invention provides a method for solving the deformation of a high-speed railway tunnel section without measurement marks, comprising: obtaining point cloud data of the high-speed railway tunnel, and aligning the point cloud data with the reference point cloud data; performing section registration on the point cloud data and the reference point cloud data; establishing a correspondence between the two phases of point cloud data through the nearest neighbor point of any point in the reference point cloud data, and setting a displacement difference threshold and an intensity grading threshold of the point pair based on the correspondence; wherein the displacement difference threshold represents that the deformation amplitude of the structure of the high-speed railway tunnel in three-dimensional space exceeds the allowable reference error, and the intensity grading threshold represents that the property change of the surface material of the high-speed railway tunnel exceeds the variation range allowed by the inherent properties of the material; clustering analysis is performed on the point clouds exceeding the displacement difference threshold or the intensity grading threshold based on the target clustering algorithm to generate a continuous mask of potential deformation areas; the continuous mask of potential deformation areas represents the area where the high-speed railway tunnel has defects.

[0006] According to one embodiment of the present invention, if the laser scanner is integrated with an inertial measurement unit, the acquisition of point cloud data of the high-speed railway tunnel and the mileage alignment of the point cloud data with the point cloud data of the reference period include: acquiring the point cloud data and the reference point cloud data based on the laser scanner, and constructing a carrier motion model of the laser scanner based on the motion observation data measured by the inertial measurement unit; outputting the pose sequence of the laser scanner when acquiring the point cloud data and the reference point cloud data respectively using the carrier motion model; interpolating the pose sequence using the Lagrange interpolation method, synchronizing the pose sequence with the scanning time of the laser scanner, and outputting the cross-sectional space parameters of the point cloud data and the reference point cloud data respectively; integrating the velocity field solved by the inertial measurement unit based on the initial mileage value in the cross-sectional space parameters to generate a calibration mileage, and mileage alignment of the point cloud data and the reference point cloud data based on the calibration mileage.

[0007] According to one embodiment of the present invention, if the laser scanner is integrated with an inertial measurement unit and an odometer, the acquiring of point cloud data of a high-speed railway tunnel and the mileage alignment of the point cloud data with the point cloud data of a reference period include: acquiring the point cloud data and the reference point cloud data based on the laser scanner, and acquiring the mileage change of the laser scanner by the odometer; and mileage alignment of the point cloud data and the reference point cloud data based on the mileage change.

[0008] According to one embodiment of the present invention, if the laser scanner does not have an integrated inertial measurement unit, the point cloud data of the high-speed railway tunnel is obtained and the point cloud data is aligned with the point cloud data of the reference period in mileage, including: obtaining the point cloud data and the reference point cloud data based on the laser scanner; performing coarse mileage alignment on the point cloud data and the reference point cloud data based on calibrated columns of the same name; constructing a feature sequence with the characteristic values ​​of the sections in the columns of the same name in the mileage direction; wherein the characteristic values ​​include the coordinates of the inner wall of the high-speed railway tunnel along the mileage direction and the changes in laser intensity, manually marked and automatically detected features; performing similarity alignment on the feature sequences of the point cloud data and the reference point cloud data through a dynamic time warping algorithm, and performing mileage alignment on the point cloud data and the reference point cloud data based on the alignment results.

[0009] According to one embodiment of the present invention, the cross-sectional registration of the point cloud data and the reference point cloud data includes: performing cross-sectional registration on the point cloud data and the reference point cloud data based on an iterative nearest point algorithm; the step of performing cross-sectional registration based on the iterative nearest point algorithm includes: establishing a set of point pairs with the nearest neighboring points of the reference point cloud data for each data point in the point cloud data; solving the transformation matrix of the point pair set, and determining the rigid body transformation matrix that minimizes the sum of squared Euclidean distances between the point pairs of the point cloud data and the reference point cloud data based on a least squares optimization method of singular value decomposition; when the rigid body transformation matrix converges, controlling the optimization to terminate, and outputting the cross-sectionally registered point cloud data and the reference point cloud data.

[0010] According to one embodiment of the present invention, the method further includes: generating a structured report including the location, size, and risk level of the defect; mapping the structured report to the tunnel Revit model components; establishing a digital archive of the high-speed railway tunnel, and conducting multi-stage deformation and defect investigation, tracking, and change detection on the high-speed railway tunnel based on the digital archive.

[0011] According to one embodiment of the present invention, the target clustering algorithm is a DBSCAN clustering algorithm.

[0012] In a second aspect, the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method of the above-mentioned embodiment when executing the computer program.

[0013] Compared with the existing technology, the beneficial effects of the present application are: through the two-dimensional coupling judgment mechanism of displacement difference threshold and intensity grading threshold, the physical credibility of deformation identification can be improved, and the two can work together to effectively distinguish between real structural damage and environmental interference. Based on the continuous deformation area mask generated by the target clustering algorithm, discrete super-threshold points are converted into spatially continuous disease representation domains, which not only realizes the precise geometric positioning of the diseased area, but also eliminates the interference of isolated noise points through spatial continuity constraints. This technical path breaks through the limitations of traditional manual inspections that are insensitive to small deformations and total station measurements that have no discrimination against material changes. Therefore, it can achieve millimeter-level deformation detection accuracy and automated disease area extraction simultaneously without preset measurement marks, thereby improving the accuracy and efficiency of high-speed rail track deformation detection, and can provide a full-section, quantitative analysis basis for high-frequency health monitoring of high-speed rail tunnels. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 Schematic diagram of the steps of the method for solving the deformation of the section of a high-speed railway tunnel without measurement marks provided in an embodiment of the present application.

[0015] Figure 2 Schematic diagram of the cross-section registration principle provided in the embodiment of the present application. DETAILED DESCRIPTION

[0016] The present application is further described in detail below in conjunction with test examples and specific implementation methods. However, this should not be understood as limiting the scope of the above-mentioned subject matter of the present application to the following embodiments. All technologies implemented based on the content of the present application fall within the scope of protection of the present application.

[0017] Unless otherwise specified, in the description of the specific embodiments of this application, the terms indicating the orientation or position relationship such as "up", "down", "left", "right", "center", "inside", "outside", and "side" are based on the expression of the orientation or position relationship shown in the accompanying drawings, or the orientation or position relationship in which the product / device / apparatus is placed when it is usually used. These terms of orientation or position relationship are only for the convenience of describing the scheme of this application or simplifying the description in the specific embodiments to facilitate the technicians to quickly understand the scheme, and do not indicate or imply that a specific device / component / element must have a specific orientation, or be constructed and operated in a specific position relationship, and therefore should not be understood as limiting this application.

[0018] In the description of the embodiments of this application, the technical terms "first," "second," etc., merely distinguish one entity or operation from another and are not to be understood as indicating or implying relative importance or implicitly specifying the quantity, specific order, or primary-secondary relationship of the indicated technical features. In the description of the embodiments of this application, "plurality" means two or more, unless otherwise specifically defined.

[0019] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0020] Please see Figure 1 , Figure 1 Schematic diagram of the steps of the method for calculating deformation of a section of a high-speed railway tunnel without measurement marks provided in an embodiment of the present application. The steps of the method for calculating deformation of a section of a high-speed railway tunnel without measurement marks may include:

[0021] S1. Obtain point cloud data of the high-speed railway tunnel using a laser scanner and align the point cloud data with the benchmark point cloud data.

[0022] In the embodiment of the present application, a laser scanner may be used to acquire point cloud data of a high-speed railway tunnel. The laser scanner may be a two-dimensional laser scanner. The point cloud data is point cloud data acquired during the current deformation detection of the high-speed railway tunnel. The reference point cloud data may be point cloud data acquired during a previous detection, or may be point cloud data acquired at a set reference time.

[0023] Since the collection process requires the use of a track-carrying trolley equipped with a laser scanner to implement multi-period data collection, and the track-carrying trolley cannot move in a completely uniform straight line during mobile scanning, there will inevitably be speed fluctuations and trajectory deviations.

[0024] Tunnel cross-sections are three-dimensional curved surfaces. Speed ​​fluctuations and trajectory deviations of the track-carrying trolley can cause variations in the laser scanner's pitch and roll angles, leading to distortion of the cross-section point cloud or deviations in normal distances. Therefore, odometry alignment of the point cloud data with the reference point cloud data is necessary to eliminate extrinsic parameter errors introduced by the track-carrying trolley's motion.

[0025] In some embodiments, if an inertial measurement unit (IMU) is integrated into the laser scanner, the mileage alignment step may include:

[0026] Based on the point cloud data and reference point cloud data acquired by the laser scanner, and the motion observation data measured by the inertial measurement unit, a carrier motion model of the laser scanner is constructed; the carrier motion model is used to output the pose sequence of the laser scanner when acquiring the point cloud data and the reference point cloud data respectively; the pose sequence is interpolated using the Lagrange interpolation method, the pose sequence is synchronized with the scanning time of the laser scanner, and the cross-sectional space parameters of the point cloud data and the reference point cloud data are output respectively; based on the initial mileage value in the cross-sectional space parameter, the velocity field solved by the inertial measurement unit is integrated to generate the calibration mileage, and the point cloud data and the reference point cloud data are aligned based on the calibration mileage.

[0027] The motion observation data measured by the IMU can include three-axis acceleration and three-axis angular velocity. A Kalman filter algorithm is used to construct a motion model of the track vehicle, which outputs a pose sequence consisting of global coordinate position and attitude quaternion. These steps eliminate IMU bias errors and convert sensor data into high-precision spatiotemporal trajectories.

[0028] Furthermore, because the IMU sampling frequency differs from the laser scanning frequency, there are some scan times at which direct pose data is not available. Therefore, Lagrange interpolation is used to interpolate the pose sequence. By interpolating the global coordinate system position and attitude quaternions, the pose sequence is synchronized with the laser scanner's scanning time, resulting in the output of cross-sectional space parameters, including the cross-sectional center coordinates, cross-sectional normal vectors, and preliminary mileage values ​​based on the IMU position integral. Based on the initial mileage values ​​in the cross-sectional space parameters, calibration mileage is integrated to generate the mileage correction for each cross-sectional area. By tangentially translating the cross-sectional point cloud along the high-speed rail track by the mileage correction value, mileage-aligned point cloud data and reference point cloud data are generated.

[0029] In other embodiments, if the laser scanner is integrated with both an inertial measurement unit and a high-precision odometer, the odometer alignment step may include:

[0030] Point cloud data and reference point cloud data are acquired based on a laser scanner, and the mileage change of the laser scanner is acquired by an odometer; and mileage alignment is performed on the point cloud data and the reference point cloud data based on the mileage change.

[0031] Among them, the odometer-based mileage alignment solution establishes a unified spatial reference by directly utilizing the displacement pulse information of the track carrier trolley. Its technical implementation is essentially a systematic mapping that converts mechanical displacement into absolute mileage coordinates. The high-precision odometer equipped with the track carrier trolley records the number of wheel rotations through an encoder and outputs a pulse signal. Each pulse corresponds to a fixed displacement (called a pulse equivalent). The displacement increment is calculated by multiplying the number of pulses by the pulse equivalent. This displacement increment is the actual distance the carrier moves along the extension direction of the track.

[0032] At the tunnel entrance, the scanning starting point is forcibly aligned with the designed mileage stake and assigned an initial mileage value. The mileage value of the cross-section point cloud collected by the laser scanner at a certain moment is determined by superimposing the displacement increment on the initial mileage. Cross-sections with the same mileage value are extracted from the benchmark point cloud data and the current point cloud data, establishing a mileage-cross-section mapping relationship. This allows the point cloud data to be aligned with the benchmark point cloud data.

[0033] In some other embodiments, if the laser scanner does not have an integrated inertial measurement unit, the odometry alignment step may include:

[0034] Point cloud data and benchmark point cloud data are acquired based on a laser scanner; coarse mileage registration of the point cloud data and benchmark point cloud data is performed based on calibrated columns of the same name; a feature sequence is constructed using the eigenvalues ​​of the sections in the columns of the same name in the mileage direction; similarity registration of the feature sequences of the point cloud data and benchmark point cloud data is performed using a dynamic time warping algorithm, and mileage alignment of the point cloud data and benchmark point cloud data is performed based on the registration results.

[0035] The eigenvalues ​​include the coordinates of the inner wall of a high-speed rail tunnel and changes in laser intensity along the mileage direction, as well as manually annotated and automatically detected features. Without other mileage positioning devices, mileage alignment is achieved through manual calibration and algorithmic automatic refinement. Laser intensity images are generated for two operations and matched in the column direction (mileage direction). First, some columns with the same name are manually calibrated through interaction to achieve coarse registration between mileages.

[0036] The sections between two identically named columns are refined and aligned, and a feature sequence is constructed using the characteristic measurements of the sections along the mileage direction. These feature values ​​include the coordinates of the tunnel wall along the mileage direction, changes in laser intensity, and features such as manually annotated or automatically detected features. The Dynamic Time Warping (DTW) algorithm is used to achieve similarity alignment between the two sequences, thereby aligning sections at the same mileage between the two phases of data. The DTW algorithm calculates the similarity between the two sequences through dynamic programming and finds the optimal warping path. This algorithm effectively eliminates sequence offsets caused by carrier velocity fluctuations or scanning errors, maximizing the similarity between the two feature sequences.

[0037] S2. Perform cross-section registration on the point cloud data and the reference point cloud data.

[0038] Please see Figure 2 , Figure 2Schematic diagram of the cross-sectional registration principle provided in the embodiment of the present application. The schematic diagram of the cross-sectional registration principle consists of multiple views, which can show the difference in effect before and after point cloud registration. The upper left corner of the figure is a verification of the local registration effect. In the interlaced part of the red and green lines, the red lines represent the reference point cloud data, the green lines represent the current point cloud data, and the line overlap reflects the registration accuracy. The lower left side of the figure is a schematic list of point cloud data parameters. The main view is a three-dimensional display of the registration results. The right side is the deformation distribution analysis of the high-speed railway tunnel. The black curve is the theoretical elevation of the tunnel design axis, and the red and green curves are the point cloud elevation distribution before and after registration, respectively.

[0039] After completing the baseline alignment of the two phases of point cloud data along the mileage direction, it is necessary to further achieve spatial alignment of the cross-section data at the same mileage to eliminate deformation errors caused by factors such as fluctuations in the inspection vehicle's posture (such as lateral shaking and pitch angle changes) during the data acquisition process.

[0040] For example, the step of performing cross-section registration may include:

[0041] The point cloud data and the reference point cloud data are cross-sectionally registered based on the iterative nearest point algorithm; the steps of cross-sectional registration based on the iterative nearest point algorithm include: establishing a set of point pairs with the nearest neighboring points of the reference point cloud data for each data point in the point cloud data; solving the transformation matrix of the point pair set, and determining the rigid body transformation matrix that minimizes the sum of squared Euclidean distances between the point pairs of the point cloud data and the reference point cloud data based on the least squares optimization method of singular value decomposition; when the rigid body transformation matrix converges, the optimization is controlled to terminate, and the cross-sectionally registered point cloud data and the reference point cloud data are output.

[0042] Among them, in view of the fact that there is only a small range of offset between the two phases of data, the iterative closest point algorithm (ICP) is used in this application to perform robust estimation of the rigid body transformation parameters, and the initial deviation can be directly set to 0.

[0043] The core process of the ICP algorithm can be decomposed into three iterative optimization stages: first, a corresponding point search is performed to establish a mapping relationship between each data point in the source point cloud and the nearest neighbor point in the target point cloud. Considering the sequential rows generated by the cross-section point cloud, only points within the range near the index number need to be matched. At the same time, a maximum matching distance threshold is set to eliminate interference from outliers.

[0044] Secondly, the transformation matrix is ​​solved. Based on the least squares optimization method of singular value decomposition (SVD), the rigid body transformation matrix that minimizes the sum of squares of Euclidean distances between corresponding point pairs can be calculated.

[0045] Finally, a convergence check is performed, and the optimization termination condition is controlled by the convergence criterion to effectively avoid local optimal solutions. This algorithm achieves precise alignment of the two cross-sectional point clouds in the XY plane, providing a highly reliable initial registration benchmark for subsequent coordinate difference calculation and intensity eigenvalue change detection.

[0046] S3. Establish a correspondence between the two phases of point cloud data through the nearest neighbor of any point in the reference point cloud data, and set the displacement difference threshold and intensity classification threshold of the point pair based on the correspondence.

[0047] The displacement difference threshold indicates when the deformation amplitude of the high-speed rail tunnel structure in three-dimensional space exceeds the allowable baseline error, and the intensity grading threshold indicates when the property changes of the high-speed rail tunnel surface material exceed the variation range allowed by the inherent characteristics of the material. Based on the two-phase point sets after registration, a multi-dimensional difference fusion calculation method is used to evaluate the tunnel structure status. The correspondence between the two phase point clouds is established through the nearest neighbor points. The Euclidean displacement of each point pair is calculated, and the displacement vector field is constructed to set the displacement difference threshold. The normalized intensity difference of each point pair is then calculated, and the intensity grading threshold is set based on the material reflectance characteristics.

[0048] S4. Based on the target clustering algorithm, cluster analysis is performed on the point clouds that exceed the displacement difference threshold or the intensity classification threshold to generate a continuous mask of the potential deformation area; the continuous mask of the potential deformation area represents the area where the high-speed railway tunnel has defects.

[0049] The DBSCAN (Density Based Spatial Clustering of Applications with Noise) algorithm can be used to cluster point clouds exceeding a threshold, marking them as potential deformation areas and generating a continuous mask of these areas. Potential deformation areas can reflect tunnel deformation or damage such as falling blocks, lining spalling, and water leakage. Manual identification and annotation of these areas can be combined with qualitative and semi-quantitative assessment of the damage extent, including crack width classification and spalling area estimation.

[0050] In summary, in the method for calculating deformation of high-speed railway tunnel sections without measurement marks provided in the embodiment of the present application, the physical credibility of deformation identification can be improved through the two-dimensional coupling judgment mechanism of displacement difference threshold and intensity grading threshold, and the two can effectively distinguish between real structural damage and environmental interference. Based on the continuous deformation area mask generated by the target clustering algorithm, discrete super-threshold points are converted into spatially continuous disease representation domains, which not only realizes the precise geometric positioning of the diseased area, but also eliminates the interference of isolated noise points through spatial continuity constraints. This technical path breaks through the limitations of traditional manual inspections that are insensitive to small deformations and total station measurements that have no discrimination against material changes. Therefore, it can achieve millimeter-level deformation detection accuracy and automated disease area extraction simultaneously without preset measurement marks, thereby improving the accuracy and efficiency of high-speed railway track deformation detection, and can provide a full-section, quantitative analysis basis for high-frequency health monitoring of high-speed railway tunnels.

[0051] Furthermore, the method provided in this application may also include:

[0052] Generate a structured report containing defect location, size, and risk level; map the structured report to the tunnel Revit model components; establish a digital archive of the high-speed rail tunnel, and conduct multi-stage deformation and defect investigation and change detection based on the digital archive.

[0053] The value of the rail digital twin is realized by automatically generating a structured report that includes defect location, size, and risk level, and mapping the inspection results to the tunnel's Revit model components. Mapping the inspection results to the BIM model creates a dynamic health profile of the tunnel, laying the foundation for subsequent intelligent operations and maintenance—the system can automatically determine which sections of the tunnel require priority repair.

[0054] Furthermore, by establishing a digital tunnel archive, traditional paper records are upgraded into a dynamic knowledge graph that integrates time and space. Using the Revit-BIM model as the spatial carrier, component-level mapping integrates heterogeneous data such as a millimeter-precise defect parameter library (crack coordinates, spalling area, leakage range), deformation evolution sequences (three-year crack expansion trajectory), and maintenance decision records (material usage, construction work orders). This creates a digital twin of the tunnel's entire lifecycle, from construction to operation and maintenance. This allows for multi-stage deformation and defect investigation and tracking, as well as change detection, improving tunnel safety monitoring and defect management.

[0055] Based on the same application concept, an embodiment of the present application also provides a computer device, which may include a memory, a processor, and a computer program stored in the memory and run on the processor. When the processor executes the computer program, it implements the method described in the above description.

[0056] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for calculating deformation of a high-speed railway tunnel section without measurement marks, characterized in that: include: Acquire point cloud data of the high-speed railway tunnel using a laser scanner, and perform mileage alignment between the point cloud data and the reference point cloud data; Performing cross-sectional registration on the point cloud data and the reference point cloud data; A correspondence between the two phases of point cloud data is established using the nearest neighbor of any point in the reference point cloud data, and a displacement difference threshold and an intensity grading threshold for the point pair are set based on the correspondence; wherein the displacement difference threshold indicates that the deformation amplitude of the high-speed railway tunnel structure in three-dimensional space exceeds the allowable reference error, and the intensity grading threshold indicates that the property change of the surface material of the high-speed railway tunnel exceeds the variation range allowed by the inherent characteristics of the material; Based on the target clustering algorithm, a cluster analysis is performed on the point cloud exceeding the displacement difference threshold or the intensity classification threshold to generate a continuous mask of potential deformation areas; the continuous mask of potential deformation areas represents the area where the high-speed railway tunnel has defects.

2. The method according to claim 1, characterized in that If the laser scanner is integrated with an inertial measurement unit, obtaining point cloud data of the high-speed railway tunnel and aligning the point cloud data with the point cloud data of the reference period include: Acquiring the point cloud data and the reference point cloud data based on a laser scanner, and constructing a carrier motion model of the laser scanner based on motion observation data measured by the inertial measurement unit; Outputting a pose sequence of the laser scanner when acquiring the point cloud data and the reference point cloud data respectively using the carrier motion model; interpolating the pose sequence using a Lagrange interpolation method, synchronizing the pose sequence with the scanning time of the laser scanner, and outputting cross-sectional space parameters of the point cloud data and the reference point cloud data respectively; Based on the initial mileage value in the cross-sectional space parameter, the velocity field solved by the inertial measurement unit is integrated to generate a calibration mileage, and the point cloud data and the reference point cloud data are aligned based on the calibration mileage.

3. The method according to claim 1, characterized in that If the laser scanner is integrated with an inertial measurement unit and an odometer, obtaining point cloud data of the high-speed railway tunnel and aligning the point cloud data with the point cloud data of the reference period includes: Acquire the point cloud data and the reference point cloud data based on a laser scanner, and acquire the mileage change of the laser scanner by the odometer; The point cloud data and the reference point cloud data are mileage-aligned based on the mileage change.

4. The method according to claim 1, wherein If the laser scanner does not have an integrated inertial measurement unit, obtaining point cloud data of the high-speed railway tunnel and aligning the point cloud data with the point cloud data of the reference period includes: Acquire the point cloud data and the reference point cloud data based on a laser scanner; Performing coarse mileage registration on the point cloud data and the reference point cloud data based on the calibrated columns of the same name; Constructing a feature sequence using the characteristic values ​​of the cross sections in the column with the same name in the mileage direction; wherein the characteristic values ​​include the coordinates of the inner wall of the high-speed railway tunnel along the mileage direction and the change of laser intensity, and the manually marked and automatically detected features; The point cloud data and the reference point cloud data are similarly registered with respect to their feature sequences using a dynamic time warping algorithm, and the point cloud data and the reference point cloud data are mileage aligned based on the registration result.

5. The method according to claim 1, characterized in that The performing cross-section registration on the point cloud data and the reference point cloud data includes: Performing cross-sectional registration on the point cloud data and the reference point cloud data based on an iterative closest point algorithm; The steps of cross-section registration based on iterative closest point algorithm include: Establishing, for each data point in the point cloud data, a point pair set of nearest neighboring points of the reference point cloud data; Solving the transformation matrix for the set of point pairs, and determining a rigid body transformation matrix that minimizes the sum of squared Euclidean distances between point pairs of the point cloud data and the reference point cloud data based on a least squares optimization method of singular value decomposition; When the rigid body transformation matrix converges, the optimization is controlled to terminate, and the cross-section-registered point cloud data and the reference point cloud data are output.

6. The method according to claim 1, wherein The method further comprises: Generate structured reports including disease location, size and risk level; Mapping the structured report to the tunnel Revit model components; A digital archive of the high-speed railway tunnel is established, and based on the digital archive, a multi-period deformation and disease investigation and tracking and change detection of the high-speed railway tunnel are carried out.

7. The method according to any one of claims 1 to 6, characterized in that The target clustering algorithm is the DBSCAN clustering algorithm.

8. A computer device, characterized in that: The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

Images