A high-precision identification method for interlayer separation in ballastless tracks

Abstract

The present invention relates to the technical field of ballastless tracks, and provides a high-precision identification method for interlayer separation of ballastless tracks, including: Step 1, collecting the separation damage of the track structure; Step 2, setting a detection area, a separation adjacent area, and a separation far-end area on the side of the track structure, and performing detection area division; Step 3, performing data smoothing and sparsification processing according to different partition attributes; Step 4, performing spatial surface fitting on the detection area and the separation adjacent area of the smoothed and sparsified processing through machine learning to obtain a reference surface; Step 5, consistency detection; Step 6, performing parallel calculation on the separation adjacent areas in different detection partitions of each track slab to obtain the size of the separation damage set of each track slab. The present invention can detect the interlayer separation of high-speed rail ballastless tracks in a non-destructive manner, and can achieve accurate and reliable measurement of sub-millimeter gaps in the case of serious damage and foreign object attachment on the detection surface.

Description

Technical Field

[0001] The present invention relates to the technical field of ballastless tracks, and more particularly, to a high-precision identification method for interlayer separation joints of ballastless tracks. Background Art

[0002] Ballastless tracks belong to typical layered structures. Under the action of trains and complex environments, separation joints occur in ballastless tracks. The separation joints will gradually deteriorate the mechanical properties of the ballastless tracks. As the service time extends, if these damages cannot be detected in time, they often rapidly develop into larger damages, further weakening the durability and service life of the ballastless track bed. To ensure the safe operation of the line, it is necessary to comprehensively and carefully detect the track state, which requires detection technologies with higher precision, more advancedness, and faster speed.

[0003] The current separation joint detection methods have the following deficiencies:

[0004] 1) Existing separation joint detection devices mainly rely on manual pushing to move forward, with relatively low detection speed and efficiency;

[0005] 2) When the camera of the existing detection device detects the separation joint at a specified position, there is an included angle during camera shooting, resulting in blind spots and affecting the detection accuracy of the separation joint;

[0006] 3) During the process of detecting damage defects, relying on full-size point cloud line fitting calculation without performing data sparse reduction operations, the detection efficiency is low;

[0007] 4) By searching for the distance between damage boundary points as the damage size determination, it is easily affected by outlier points and lacks a reference plane;

[0008] 5) When using the distance between two intersection points of the fitted straight line as the track structure height difference characteristic value, the straight line fitting is prone to abnormal angles of the straight line;

[0009] 6) PCA data dimensionality reduction and setting filter sampling are prone to losing local information. When there are missing data, it is easily interfered by noise points;

[0010] 7) The traditional least squares method fits the surface by minimizing the sum of squared errors, but it is usually only applicable to simple linear or polynomial surfaces and is difficult to handle complex non-linear data distributions. Summary of the Invention

[0011] The content of the present invention is to provide a high-precision identification method for interlayer separation joints of ballastless tracks, which can detect the interlayer separation joints of high-speed rail ballastless tracks in a non-destructive manner and can accurately and reliably measure sub-millimeter-level gaps in the case of severe damage and foreign object attachment on the detection surface.

[0012] A high-precision identification method for interlayer separation joints of ballastless tracks according to the present invention includes the following steps:

[0013] Step 1: Collect the joint separation damage of the track structure;

[0014] Step 2: Set up a detection area, a joint separation adjacent area, and a joint separation far-end area on the side of the track structure, and conduct detection area division;

[0015] Step 3: Perform data smoothing and sparsification processing according to different partition attributes;

[0016] Step 4: Through machine learning, enhance the spatial surface fitting of the detection area and the joint separation adjacent area after smoothing and sparsification processing to obtain a reference surface;

[0017] Step 5: Consistency detection;

[0018] Step 6: Conduct parallel calculations in the joint separation adjacent areas within different detection partitions of each track slab to obtain the size of the joint separation damage set for each track slab.

[0019] Preferably, in Step 1, the damage of the joint separation of the track structure is collected by a hub encoder, a line laser 3D stereo camera, and a two-axis robotic arm, specifically as follows:

[0020] 1.1) Expand the two-axis robotic arm, and automatically adjust the line laser 3D stereo camera to vertically photograph the damage position at the track slab and the adjustment layer according to the corresponding track type. When there are interference objects on the side of the track, the robotic arm automatically lifts up and continues to collect after passing through the interference objects;

[0021] 1.2) Enable the hub encoder, and drive the laser 3D stereo camera into the acquisition working state in the external pulse drive mode;

[0022] 1.3) Start the vehicle and continuously collect and store the joint separation point cloud depth image data at high speed.

[0023] Preferably, in Step 2, the detection area, the joint separation adjacent area, and the joint separation far-end area are all rectangular areas. The detection area completely covers the joint separation, and the side of the track structure is set as the detection area; the horizontal midline of the joint separation adjacent area coincides with the horizontal midline of the interlayer joint separation; the ratio of the height of the joint separation adjacent area to the height of the detection area is n, and the value range of n is 10% - 20%; within the detection area, after removing the joint separation adjacent area, the remaining area is the joint separation far-end area.

[0024] Preferably, in Step 3, specifically:

[0025] 3.1) For the joint separation adjacent area, retain all data or perform downsampling by taking every other one. For the joint separation adjacent area in the i-th detection partition, the data set is denoted as D i , i = 0, 1, 2,... m - 1;

[0026] 3.2) For the joint separation far-end area, perform downsampling by taking every a-th one, where a takes an integer value between 3 and 7;

[0027] 3.3) Denote the data set belonging to the $i$-th detection partition as $P$ i , where $i = 0, 1, 2, \ldots, m - 1$;

[0028] 3.4) Denote the data set of the distal region of the seam separation belonging to the $i$-th detection partition as $P$ Li , where $i = 0, 1, 2, \ldots, m - 1$;

[0029] 3.5) In the data set $P$ i , the point coordinate data format in its three-dimensional space is $(x$ b , $y$ i , $z$ i ). Then, for $m$ data, define the following piecewise function $C(j\Delta y)$:

[0030]

[0031] In the formula, $j$ takes values $0, 1, 2, \ldots, m - 1$, $\Delta y$ is the discrete step size; $y_0, y_1, y_2, y$ m-1 represent the $y$-direction coordinates, and $z_0, z_1, z$ m-1 represent the $z$-direction coordinates;

[0032] Set the discrete step size:

[0033]

[0034] Define the control variable $e$ λ (j):

[0035] $e$ λ (j) = $C(j\Delta y) - \text{Pos}(j) + \lambda\|\text{Pos}(j)\|_1$

[0036] In the formula, $\text{POS}(j)$ is a smoothing function, and $\lambda > 0$ is the sparse regularization coefficient;

[0037] Sparse smoothing function iteration:

[0038] $\text{Pos}(j + 1) = \text{Pos}(j) + f$ λ (j)

[0039] In the formula, $f$ λ (j) is a smoothing factor that includes a combination of the control variable and weights;

[0040]

[0041] In the formula, $s_0, s_1, s_2$ are coefficients for adjusting the smoothing degree;

[0042] Define the initial iteration conditions:

[0043]

[0044] e(0) = e(-1) = 0

[0045] After iterative calculation, the following data set is finally obtained:

[0046] (x b , y0, Pos(0)), (x b , y0 + Δy, Pos(1)), (x b , y0 + 2Δy, Pos(2)),..., (x b , y0 + jΔy, Pos(j))

[0047] 3.6) Replace P with the smoothed and sparse processed data obtained above i data set.

[0048] Preferably, in step 4, specifically:[[]]

[0049] 4.1) Perform spatial surface fitting on the data sets of the first detection partition, the middle detection partition, and the last detection partition of each track slab;

[0050] 4.2) Construct a feedforward neural network, and the network output z represents the surface function of the detection partition data set, that is, z = f(x, y, θ);

[0051] 4.3) Optimize the target surface by combining the least square error and regularization, update the machine learning model parameter θ using the gradient descent method, and use the optimal solution of the least square method as the initial weight;

[0052] 4.4) Initialize the model parameter θ, obtain a preliminary fitting result through the least square method as the initial value of the machine learning model, perform iterative optimization using the training data, and minimize the loss function θ * , and adjust the regularization coefficient λ' according to the validation set error;

[0053]

[0054] In the formula, R(θ) represents regularization, and N represents the total number of discrete points;

[0055] 4.5) When the loss drops to the minimum, output the fitting surface function.

[0056] Preferably, in step 5, specifically:[[]]

[0057] 5.1) In the spatial surface fitting functions of the first detection partition, the middle detection partition, and the last detection partition of each track slab, the extracted surface coefficient vectors are V0, V (l-1) / 2 and V l-1 ;

[0058] 5.2) Calculate V0-V (l-1) / 2 , V0-V l-1 , V (l-1) / 2 -V l-1 's l2 norm, and calculate the variance var of these 3 norms;

[0059] 5.3) If the variance var is less than or equal to the preset threshold th, and the value range of the threshold th is 2-10, it is considered that the fitting surface has good consistency, that is, the fitting reference surface of the detection partition coincides with the reference surface of the global fitting of each track side surface in height;

[0060] 5.4) Use the average value of the fitting surfaces of the three detection partitions as the global reference surface.

[0061] Preferably, in step 6, specifically:

[0062] 6.1) In the i-th detection partition, the reference surface is uniformly denoted as basep i , i = 0, 1, 2, 3…m-1;

[0063] 6.2) In the data set P Li of the adjacent area of the gap, all the points located below basep i are used as the points inside the gap, and the points with the same x coordinate among them are successively denoted as: col_low_i, i = 0, 1, 2, 3…, similarly, the set of points above and on the basep i surface is denoted as: col_high_i, i = 0, 1, 2, 3…; and the points in this set are sorted in ascending order based on their y coordinates;

[0064] 6.3) Traverse each data set col_low_i, calculate the distance from each point in it to basep i and record the result dis_i, i = 0, 1, 2, 3…, find and record the maximum distance value as d_max_i in dis_i;

[0065] 6.4) If d_max_i is greater than or equal to the preset gap detection depth threshold, and the value range of the depth threshold is 5mm-10mm, it is determined that there is a gap damage. In dis_i, take the average value of the q points closest to basep i as the gap depth, and the value range of q is 3-5;

[0066] 6.5) Calculate the 2 points with the farthest adjacent distance in the x direction among the set of points. These 2 points are the edge points of the gap at this position, and the direct distance between them is the width of the gap. If col_low_i is an empty set, use col_high_i to determine the width of the gap at this position;

[0067] 6.6) Count the number k of positions continuously determined as off-seam positions. Each detected position indicates that there is an actual off-seam of c millimeters in length. The total length of the off-seam in this data frame is k * c;

[0068] 6.7) Aggregate and statistically analyze the results after parallel computing to obtain the size of the off-seam damage set for each track slab.

[0069] The beneficial effects of the present invention are as follows:

[0070] (1) The technical solution of using a wheel hub encoder + a two-axis robotic arm + a 3D stereo camera enables the camera to collect 3D data with the least distortion perpendicular to the measured surface. When collecting data, the wheel hub encoder outputs sampling pulses based on the rotation angle of the wheel, is insensitive to the vehicle speed uniformity, and by accumulating and counting the sampling pulses, the vehicle driving distance can be converted, and the geographical information of data sampling can be located simultaneously;

[0071] (2) In the processing method, the concepts of off-seam adjacent area, off-seam far-end area, and detection area are proposed, and the ranges and downsampling parameters of each area are defined. In the off-seam far-end area, sparse sampling is carried out to a large extent. Through a relatively small amount of data and relying on the breadth of its spatial distribution, the accuracy of the fitting surface is ensured. The above measures effectively reduce the time complexity and space complexity of the method;

[0072] (3) Using the surface model given by the present invention as the fitting target of the reference plane, even when there is a surface deformation on the measured surface, it can still closely approximate the real surface with high precision. At the same time, if the measured surface is a high-quality plane, this model can also degenerate into a spatial plane model to meet the fitting accuracy requirements;

[0073] (4) The concept of parallel computing for detection sub-areas is proposed, and the number is set to be odd. The size of the off-seam damage set is calculated in parallel in different detection sub-areas, which speeds up the calculation speed. Brief Description of the Drawings

[0074] Figure 1 It is a flowchart of a high-precision identification method for interlayer off-seams of a ballastless track in an embodiment;

[0075] Figure 2 It is a front view of the data collection process of interlayer off-seams of a ballastless track in an embodiment;

[0076] Figure 3 It is a side view of the data collection process of interlayer off-seams of a ballastless track in an embodiment;

[0077] Figure 4 It is the division of the off-seam adjacent area, off-seam far-end area, and detection sub-areas of interlayer off-seams of a ballastless track in an embodiment;

[0078] Figure 5The numbering of the side detection partitions for each track slab in the embodiment. Detailed implementation manners

[0079] To further understand the content of the present invention, the present invention will be described in detail in combination with the accompanying drawings and embodiments. It should be understood that the embodiments are only for explaining the present invention rather than limiting it.

[0080] Embodiment

[0081] As Figure 1 shown, this embodiment provides a high-precision recognition method for the interlayer separation of the ballastless track, which includes the following steps:

[0082] Step 1: Collect the separation damage of the track structure;

[0083] Step 2: Set a detection area, a separation adjacent area, and a separation far-end area on the side of the track structure, and conduct detection area division;

[0084] Step 3: Perform data smoothing and sparsification processing according to different partition attributes;

[0085] Step 4: Through machine learning, enhance the spatial surface fitting of the detection area and the separation adjacent area after smoothing and sparsification processing to obtain a reference reference surface;

[0086] Step 5: Consistency detection;

[0087] Step 6: Perform parallel calculation in the separation adjacent area within different detection partitions of each track slab to obtain the size of the separation damage set of each track slab.

[0088] In step 1, the damage of the track structure separation 3 is collected through the hub encoder 1, the line laser 3D stereo camera 2, and the dual-axis robotic arm 4, as Figure 2 and Figure 3 shown, specifically as follows:

[0089] 1.1) Expand the dual-axis robotic arm 4, and automatically adjust the line laser 3D stereo camera (2) to the vertical shooting damage position at the track slab and the adjustment layer according to the corresponding track type. When there are interference objects on the track side, the robotic arm automatically lifts and continues to collect after passing through the interference objects;

[0090] 1.2) Enable the hub encoder 1, and drive the laser 3D stereo camera 2 into the acquisition working state in the external pulse drive mode;

[0091] 1.3) Start the vehicle and continuously collect and store the separation point cloud depth image data at high speed.

[0092] In step 2, as Figure 4As shown, the detection area 7, the adjacent area 5 to the crack, and the far - end area 6 to the crack are all rectangular areas. The detection area completely covers the crack (or the area where the crack may occur), and the side of the track structure is set as the detection area; the horizontal mid - line of the adjacent area to the crack coincides with the horizontal mid - line of the inter - layer crack; the ratio of the height of the adjacent area to the crack to the height of the detection area is n, and the value range of n is 10% - 20%; within the detection area, after removing the adjacent area to the crack, the remaining area is the far - end area to the crack.

[0093] Each track slab is evenly divided into l rectangular areas, where l must be an odd number, and the recommended value is 7. Parallel calculations are performed in different detection areas. The specific form is as Figure 5 shown. The detection sub - areas are numbered 0, 1, 2, 3... from left to right.

[0094] In step 3, specifically:

[0095] 3.1) For the adjacent area to the crack, retain all data or perform down - sampling by taking every other one. For the adjacent area to the crack in the i - th detection sub - area, the data set is denoted as D i , i = 0, 1, 2,... m - 1;

[0096] 3.2) For the far - end area to the crack, perform down - sampling by taking every a one, where a takes an integer value between 3 and 7;

[0097] 3.3) For the data belonging to the i - th detection sub - area, the data set is denoted as P i , i = 0, 1, 2,... m - 1;

[0098] 3.4) For the far - end area to the crack belonging to the i - th detection sub - area, the data set is denoted as P Li , i = 0, 1, 2,... m - 1;

[0099] 3.5) In the data set P i , the point coordinate data format in its three - dimensional space is (x b , y i , z i ). Then, for m data, the following piece - wise function C(jΔy) is defined:

[0100]

[0101] In the formula, j takes values 0, 1, 2... m - 1, Δy is the discrete step; y0, y1, y2, y m-1 represent the y - direction coordinates, and z0, z1, z m-1 represent the z - direction coordinates;

[0102] Set the discrete step:

[0103]

[0104] Define the control variable eλ (j):

[0105] e λ (j) = C(jΔy) - Pos(j) + λ||Pos(j)||1

[0106] where POS(j) is a smoothing function and λ > 0 is the sparse regularization coefficient;

[0107] Sparse smoothing function iteration:

[0108] Pos(j + 1) = Pos(j) + f λ (j)

[0109] where f λ (j) is a smoothing factor that includes a combination of control variables and weights;

[0110]

[0111] where s0, s1, and s2 are coefficients for adjusting the smoothing degree;

[0112] Define the initial iteration conditions:

[0113]

[0114] e(0) = e(-1) = 0

[0115] After iterative calculation, the following data set is finally obtained:

[0116] (x b , y0, Pos(0)), (x b , y0 + Δy, Pos(1)), (x b , y0 + 2Δy, Pos(2)),..., (x b , y0 + jΔy, Pos(j))

[0117] 3.6) Replace P with the smoothed and sparse processed data obtained above i Data set.

[0118] In step 4, specifically:

[0119] 4.1) Perform spatial surface fitting on the data sets of the first detection partition, the middle detection partition, and the last detection partition of each track slab;

[0120] 4.2) Construct a feedforward neural network (with a network structure of more than three layers), and the network output z represents the surface function of the detection partition data set, that is, z = f(x, y, θ);

[0121] 4.3) Optimize the target surface by combining the least squares error and regularization, update the machine learning model parameter θ using the gradient descent method, and use the optimal solution of the least squares method as the initial weight;

[0122] 4.4) Initialize the model parameter θ, obtain the preliminary fitting result through the least squares method as the initialization value of the machine learning model, and use the training data for iterative optimization to minimize the loss function θ * , and adjust the regularization coefficient λ' according to the validation set error;

[0123]

[0124] In the formula, R(θ) represents regularization, and N represents the total number of discrete points;

[0125] 4.5) When the loss drops to the minimum, output the fitting surface function.

[0126] In step 5, specifically:

[0127] 5.1) In the spatial surface fitting functions of the first detection partition, the middle detection partition, and the last detection partition of each track slab, the extracted surface coefficient vectors are V0, V (l-1) / 2 and V l-1 ;

[0128] 5.2) Calculate the l2 norms of V0 - V (l-1) / 2 , V0 - V l-1 , V (l-1) / 2 -V l-1 respectively, and calculate the variance var of these three norms;

[0129] 5.3) If the variance var is less than or equal to the preset threshold th (th is recommended to take a value between 2 and 10), it is considered that the fitting surface has good consistency, that is, the fitting reference surface of the detection partition coincides with the reference surface of the global fitting of each track side;

[0130] 5.4) Use the average value of the fitting surfaces of the three detection partitions as the global reference surface.

[0131] In step 6, specifically:

[0132] 6.1) In the i-th detection partition, the reference surface is uniformly denoted as basep i , i = 0, 1, 2, 3…m - 1;

[0133] 6.2) In the off-seam adjacent area data set P Li , all data located in basep iThe points below, as the points within the gap, and the points with the same x coordinate among them are sequentially denoted as: col_low_i, i = 0, 1, 2, 3… Similarly, the set of points above and on the basep i plane is denoted as: col_high_i, i = 0, 1, 2, 3…; and the points in this set are sorted in ascending order based on their y coordinates;

[0134] 6.3) Traverse each data set col_low_i, calculate the distance from each point in it to basep i and record the result dis_i, i = 0, 1, 2, 3…, find and record the maximum distance value as d_max_i in dis_i;

[0135] 6.4) If d_max_i is greater than or equal to the preset gap detection depth threshold (the depth threshold is recommended to take a value between 5 mm and 10 mm), it is determined that there is a gap damage. In dis_i, take the average value of the q points closest to basep i as the gap depth, and the value range of q is 3 - 5;

[0136] 6.5) Calculate the 2 points with the farthest adjacent distance in the x direction among the set of points. These 2 points are the edge points of the gap here, and the direct distance between them is the width of the gap. If col_low_i is an empty set, use col_high_i to determine the gap width at this position;

[0137] 6.6) Count the number k of consecutive positions determined as gaps. Each time 1 gap is detected, it indicates that there is actually a gap c millimeters long. k * c is the total length of the gaps in this data frame;

[0138] 6.7) Summarize and statistically calculate the results after parallel computing to obtain the size of the gap damage set for each track slab.

[0139] In this embodiment, a gap acquisition device with a wheel hub encoder + line laser 3D stereo camera + dual-axis robotic arm is designed, which can automatically adjust the laser camera to extend to the track slab and the adjustment layer position according to the track type, and vertically acquire the gap point cloud depth image to ensure the acquisition accuracy. This device can perform fast line array high-precision acquisition, with a resolution of 2560 px and a scanning frame rate of 46 khz, ensuring an acquisition and recognition accuracy of 0.1 mm at a speed of 16.5 km / h.

[0140] To increase the detection efficiency and reduce the recognition complexity, this embodiment proposes a method for sparse dimensionality reduction of partition data of the gap damage point cloud depth image, sets different partitions in the point cloud depth image, and reduces the data capacity of the non-gap area.

[0141] Due to the differences in the construction or casting processes of the track slab and the adjustment layer, there are slight arcs on the contact surface. To ensure high-precision recognition, this embodiment proposes a high-order surface reference for joint gap recognition to reduce the accuracy impact caused by outliers or slight arcs.

[0142] The above has schematically described the present invention and its implementation manners. This description is not restrictive, and what is shown in the drawings is only one of the implementation manners of the present invention. The actual structure is not limited thereto. Therefore, if those of ordinary skill in the art are inspired by it and, without departing from the gist of the present invention, creatively design structural manners and embodiments similar to this technical solution, they shall fall within the protection scope of the present invention.

Claims

1. A high-precision method for identifying inter-layer gaps in ballastless track, characterized by: The following steps are involved: Step 1: Collect track structure damage; Step 2: Set up a detection area, a gap adjacent area, and a gap distal area on the side of the track structure, and divide the detection areas; Step 3: Perform data smoothing and sparse processing according to different partition attributes; Step 4: Perform spatial surface fitting on the smoothed and sparsely processed detection area and the adjacent area of the gap through machine learning enhancement to obtain a reference surface; In step 4, specifically: 4.1) Perform spatial surface fitting on the data sets of the first, middle, and last inspection partitions of each track plate; 4.2) Construct a feedforward neural network, where the network output z represents the surface function of the detection partition data set, i.e. z = f(x, y, θ); 4.3) Optimize the target surface by combining the least squares error and regularization, and use the gradient descent method to update the machine learning model parameters θ, while using the optimal solution of the least squares method as the initial weights; 4.4) Initialize the model parameters θ and obtain preliminary fitting results through the least squares method. Use them as the initialization values of the machine learning model and perform iterative optimization using the training data to minimize the loss function θ. * , adjust the regularization coefficient λ' according to the validation set error; In the formula, R(θ) represents regularization, and N represents the total number of discrete points; 4.5) When the loss drops to a minimum, output the fitted surface function; Step 5: Consistency test; In step 5, specifically: 5.1) In the spatial surface fitting function of the first detection partition, the middle detection partition, and the last detection partition of each track plate, the extracted surface coefficient vectors are V0, V (l-1) / 2 and V l-1 ; 5.2) Calculate V0-V respectively (l-1) / 2 , V0-V l-1 , V (l-1) / 2 -V l-1 The l2 norm of , and calculate the variance var of this 3 norms; 5.3) If the variance var is less than or equal to the preset threshold th, where the threshold th ranges from 2 to 10, the fitted surface is considered to have good consistency, i.e., the fitted reference surface of the detection partition is highly consistent with the reference surface of the global fitting of each track side; 5.4) The average value of the fitted surfaces of the three detection partitions is used as the global reference surface; Step 6: Perform parallel calculations on the adjacent areas of the gaps in different inspection zones of each track slab to obtain the aggregate size of the gap damage on each track slab.

2. The high-precision method for identifying inter-layer gaps in ballastless track according to claim 1 is characterized in that: In step 1, the damage of the track structure gap (3) is collected through the wheel hub encoder (1), the line laser 3D stereo camera (2), and the dual-axis robotic arm (4), as follows: 1.1) Deploy the dual-axis robotic arm (4) and automatically adjust the line laser 3D stereo camera (2) to the track plate and the adjustment layer to vertically capture the damage location according to the corresponding track type. When there is an interference on the side of the track, the robotic arm automatically lifts up and passes through the interference to continue collecting; 1.2) Enable the wheel hub encoder (1), and use the external pulse drive mode to drive the laser 3D stereo camera (2) to enter the acquisition working state; 1.3) Start the vehicle and continuously collect and store the off-seam point cloud depth image data at high speed.

3. The high-precision method for identifying inter-layer gaps in ballastless track according to claim 2, characterized in that: In step 2, the detection area, the area adjacent to the gap, and the area far from the gap are all rectangular areas. The detection area completely covers the gap, and the side of the track structure is set as the detection area; the horizontal center line of the area adjacent to the gap coincides with the horizontal center line of the inter-layer gap; the ratio of the height of the area adjacent to the gap to the height of the detection area is n, and the value range of n is 10%-20%; in the detection area, after removing the area adjacent to the gap, the remaining area is the area far from the gap.

4. The high-precision method for identifying inter-layer gaps in ballastless track according to claim 3 is characterized in that: In step 3, specifically: 3.1) For the adjacent area of the gap, all data are retained, or every other 1 downsampling is performed. The data set of the adjacent area of the gap of the i-th detection partition is recorded as D i , i=0,1,2,…m-1; 3.2) In the area far from the seam, downsampling is performed every a, where a is an integer between 3 and 7; 3.3) The data set belonging to the i-th detection partition is recorded as P i , i=0,1,2,…m-1; 3.4) The data set of the remote area from the crack belonging to the i-th detection partition is recorded as P Li , i=0,1,2,…m-1; 3.5) In P i In the data set, the point coordinate data format of the three-dimensional space is (x b ,y i ,z i ), then for m data, define the following piecewise function C(jΔy): Where, j takes the value 0, 1, 2...m-1, Δy is the discrete step length; y0, y1, y2, y m-1 Indicates the y-direction coordinate, z0, z1, z m-1 Indicates the z-direction coordinate; Set discrete steps: Define the control quantity e λ (j): e λ (j)=C(jΔy)-Pos(j)+λ||Pos(j)||1 Where POS(j) is a smoothing function, λ>0 is a sparse regularization coefficient; Sparse smooth function iteration: Pos(j+1)=Pos(j)+f λ (j) Where, f λ (j) is a smoothing factor, which includes a combination of control amount and weight; Where s0, s1, and s2 are coefficients for adjusting the degree of smoothness; Define the initial conditions for the iteration: After iterative calculation, the following data set is finally obtained: (x b ,y0,Pos(0)),(x b ,y0+Δy,Pos(1)),(x b ,y0+2Δy,Pos(2)),...,(x b ,y0+jΔy,Pos(j)) 3.6) Replace P with the smoothed sparse data obtained above i Data collection.

5. The high-precision method for identifying inter-layer gaps in ballastless track according to claim 4 is characterized in that: In step 6, specifically: 6.1) For the i-th detection zone, the reference surface is uniformly recorded as basep i ,i=0,1,2,3…m-1; 6.2) Set the adjacent area data set P Li In basep i The points below are regarded as points in the gap, and the points with the same x coordinate are recorded as col_low_i, i = 0, 1, 2, 3... Similarly, all points above and at basep are recorded as col_low_i, i = 0, 1, 2, 3... i The set of points on the surface is denoted as col_high_i, where i = 0, 1, 2, 3, etc. The points in this set are arranged in ascending order based on their y coordinates. 6.3) Traverse each data set col_low_i and calculate the distance from each point to basep i The distance is recorded and the result is dis_i, i = 0, 1, 2, 3..., and the maximum distance value is found in dis_i and recorded as d_max_i; 6.4) If d_max_i is greater than or equal to the preset crack detection depth threshold, the depth threshold value range is 5mm-10mm, then it is determined that there is crack damage. In dis_i, take the distance basep i The average value of the nearest q points is the crack depth, and the value of q ranges from 3 to 5; 6.5) Calculate the two points in the set that are the farthest apart in the x-direction. These two points are the edge points of the gap. The distance between them is the gap width. If col_low_i is an empty set, use col_high_i to determine the gap width at this location. 6.6) Count the number k of consecutive crack positions determined to be cracks. Each time one is detected, it indicates that there is actually a crack of c mm in length. k*c is the total length of the cracks in this data frame. 6.7) Summarize the results of parallel calculations to obtain the aggregate size of the gap damage of each track slab.

Images