Dynamic measurement method and device for turnout point rail lowering value

Abstract

The application discloses a kind of dynamic measurement method and device of turnout point rail lowering value, wherein the method includes: using pre-trained multi-scale semantic segmentation model to classify and identify turnout point cloud data, output the point cloud data of basic rail and multiple turnout structure components;The turnout structure component includes wing rail, point rail and heart rail;The multi-scale semantic segmentation model is based on PointNet++ training obtains;Based on least square method, respectively using the point cloud data of basic rail and wing rail is fitted to obtain rail top surface curve, rail head side curve, obtains rail top point, gauge point;According to the geometric structure relationship of basic rail and point rail, based on rail head side curve determines lowering value feature point;Using the height difference of rail top point and lowering value feature point, determine point rail lowering value, fixed heart rail lowering value.The application can improve the accuracy and efficiency of turnout point rail lowering value measurement.

Description

Technical Field

[0001] This invention relates to the field of railway infrastructure technology, and in particular to a dynamic measurement method and device for the drop value of the turnout tip rail. Background Technology

[0002] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section.

[0003] As a critical steering device in the track system, the structural reliability of turnouts directly affects train operation safety. The high-frequency impact and lateral forces generated when trains pass over turnouts can easily cause wear and deformation of key components such as the switch rail and frog rail, thereby reducing the turnout's service life and even causing safety accidents. With the continuous increase in train speed and capacity, higher requirements are placed on the safety and stability of turnout systems. Among these, the drop between the switch rail and frog rail (i.e., the height difference between the top surfaces of the switch rail and frog rail and the top surface of the stock rail) is a core indicator of wheel-rail contact geometry, directly affecting wheel load transition characteristics, vehicle vibration, and operating speed, thus becoming a key control indicator for turnout design and maintenance.

[0004] In the field of turnout condition inspection, current methods mainly rely on manual inspection, which is limited by operational experience and visual measurement accuracy, making it difficult to meet the automated inspection requirements of the "Standards for Repair of Conventional Railway Lines". Although two-dimensional vision inspection technology has been applied in track inspection, its planar imaging characteristics are easily affected by changes in lighting, component occlusion, and installation posture, resulting in insufficient robustness of feature extraction in complex turnout scenarios. In contrast, three-dimensional point cloud technology can obtain spatial coordinate data containing depth information, which can more accurately describe the three-dimensional geometric features of turnout components. However, existing railway three-dimensional point cloud research mainly focuses on ordinary track structures (such as fastener and sleeper inspection), and turnout scenarios face challenges such as complex component geometry, sparse turnout point cloud data, and coupling of spatial relationships among multiple components, making it difficult for traditional point cloud segmentation algorithms to achieve accurate identification and efficient processing of turnout structures. Summary of the Invention

[0005] This invention provides a dynamic measurement method for the drop value of the turnout point rail, to improve the accuracy and efficiency of turnout point rail drop value measurement. The method includes:

[0006] Acquire turnout point cloud data;

[0007] A pre-trained multi-scale semantic segmentation model is used to classify and identify turnout point cloud data, outputting point cloud data of the base rail and multiple turnout structural components; the turnout structural components include wing rail, switch rail, and frog rail; the multi-scale semantic segmentation model is trained based on PointNet++.

[0008] Based on the least squares method, the rail top surface curve and rail head side surface curve are obtained by fitting the point cloud data of the basic rail and the wing rail respectively.

[0009] Based on the rail top surface curve and the rail head side curve, the rail vertex and gauge point are obtained;

[0010] Based on the geometric relationship between the stock rail and the switch rail, the descent feature point is determined based on the side curve of the rail head; the descent feature point reflects the apex of the switch rail or the frog rail.

[0011] The height difference between the track apex and the lowering feature point is used to determine the lowering value of the switch rail and the lowering value of the fixed center rail.

[0012] This invention also provides a dynamic measurement device for the drop value of the turnout point rail, to improve the accuracy and efficiency of the measurement of the drop value of the turnout point rail. The device includes:

[0013] The data acquisition module is used to acquire turnout point cloud data;

[0014] The segmentation model processing module is used to classify and identify turnout point cloud data using a pre-trained multi-scale semantic segmentation model, and output point cloud data of the base rail and multiple turnout structural components; the turnout structural components include wing rail, switch rail and frog rail; the multi-scale semantic segmentation model is trained based on PointNet++.

[0015] The degradation calculation module is used to obtain the rail top surface curve and rail head side curve by fitting the point cloud data of the base rail and wing rail respectively using the least squares method; based on the rail top surface curve and rail head side curve, the rail vertex and gauge point are obtained; according to the geometric relationship between the base rail and the switch rail, the degradation feature point is determined based on the rail head side curve; the degradation feature point reflects the vertex of the switch rail or the frog rail; using the height difference between the rail vertex and the degradation feature point, the degradation value of the switch rail and the fixed frog rail are determined.

[0016] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned dynamic measurement method for the drop value of the turnout tip rail.

[0017] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for dynamically measuring the drop value of the turnout tip rail.

[0018] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-mentioned dynamic measurement method for the drop value of the turnout tip rail.

[0019] In this embodiment of the invention, the PointNet++ model is first used to classify and identify the turnout point cloud data, outputting point cloud data of the basic rail and multiple turnout structural components, thus improving the identification accuracy of each component in the turnout point rail scenario. Then, an adaptive least squares fitting algorithm is used to extract the rail vertices, gauge points, and drop value feature points, achieving high-precision measurement of the point rail drop value and the fixed point rail drop value. Experimental verification shows that the method achieves a measurement accuracy of ±0.3mm under dynamic detection conditions, meeting the millimeter-level detection standard for high-speed railway turnouts, and improving the accuracy and efficiency of turnout point rail drop value measurement. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0021] Figure 1 This is a flowchart illustrating the dynamic measurement method for the drop value of the turnout tip rail in an embodiment of the present invention.

[0022] Figure 2 This is a specific example diagram of the dynamic measurement method for the drop value of the turnout tip rail in an embodiment of the present invention;

[0023] Figure 3 This is another specific example of the dynamic measurement method for the drop value of the turnout tip rail in the embodiments of the present invention;

[0024] Figure 4 This is a schematic diagram of the rail top surface curve and the rail head side surface curve in an embodiment of the present invention;

[0025] Figure 5 This is a schematic diagram of the process for calculating the reduction value of the pointed rail in an embodiment of the present invention;

[0026] Figure 6 This is a schematic diagram illustrating the calculation of the reduced value feature points in an embodiment of the present invention;

[0027] Figure 7 This is a schematic diagram of the dynamic measurement device for the reduction value of the turnout tip rail in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0029] There are two types of turnout reduction values: one is the reduction value of the switch rail relative to the stock rail, and the other is the reduction value of the frog rail relative to the wing rail. The reduction value can be obtained by measuring the vertical distance between the feature point and the top of the stock rail and the wing rail. Given the significant structural differences between the stock rail, switch rail, and frog rail, the switch rail can move to insert into or leave the gap between the stock rails; the frog rail includes movable frog rail and fixed frog rail structures: (1) When it is necessary to open a track in a certain direction, the movable frog rail frog is closely attached to the wing rail in the same direction of opening, while separating from the other wing rail, thereby eliminating the harmful space in the turnout. (2) The wing rail of the fixed frog rail is located on both sides of the frog rail body, forming a V-shaped structure together with the frog rail body [25-27]. Therefore, after obtaining the three-dimensional data of the turnout in the track reference coordinate system, due to the complex structural relationship between the switch rail and the frog rail, it is necessary to accurately extract feature points from the scan data to calculate the reduction value.

[0030] According to existing regulations, when the drop in the point rail exceeds 1mm and affects the stability of train operation, it must be repaired or replaced. To improve measurement accuracy, it is essential to accurately identify turnout components and extract feature points.

[0031] To address the technical bottlenecks in existing technologies, this invention proposes a dynamic measurement method for the drop value of the point rail based on multi-field three-dimensional vision, in conjunction with the onboard dynamic detection platform of the China Academy of Railway Sciences. First, high-density turnout point cloud data is acquired through multi-sensor calibration and field-of-view fusion technology. Second, the PointNet++ point cloud segmentation model is improved, and a turnout partitioning downsampling strategy is introduced to enhance the accuracy of component recognition in complex scenarios. Finally, an adaptive least-squares fitting algorithm is used to extract rail top feature points, achieving high-precision measurement of the drop value.

[0032] Figure 1 This is a flowchart illustrating the dynamic measurement method for the drop value of the turnout tip rail in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes:

[0033] Step 101: Obtain turnout point cloud data;

[0034] Step 102: Classify and identify the turnout point cloud data using a pre-trained multi-scale semantic segmentation model, and output the point cloud data of the basic rail and multiple turnout structural components; the turnout structural components include wing rail, switch rail and frog rail; the multi-scale semantic segmentation model is trained based on PointNet++.

[0035] Step 103: Based on the least squares method, the top surface curve and the side surface curve of the rail head are obtained by fitting the point cloud data of the basic rail and the wing rail respectively.

[0036] Step 104: Based on the rail top surface curve and the rail head side curve, obtain the rail apex and gauge point;

[0037] Step 105: Based on the geometric relationship between the base rail and the switch rail, determine the descent feature point based on the side curve of the rail head; the descent feature point reflects the apex of the switch rail or the frog rail.

[0038] Step 106: Use the height difference between the rail apex and the lowering feature point to determine the lowering value of the switch rail and the lowering value of the fixed center rail.

[0039] The following is a detailed explanation of the dynamic measurement method for the drop value of the turnout tip rail in the embodiments of the present invention.

[0040] In step 101, the turnout point cloud data is obtained.

[0041] A turnout detection system based on structured light measurement principles is used to acquire, process, and analyze three-dimensional data of the track surface and surrounding facilities. The three-dimensional turnout data includes not only the stock rail but also turnout components (switches, frog rails, wing rails, and guard rails), sleepers, fasteners, and ballast. Therefore, to avoid interference from irrelevant structures in the measurement of turnout degradation, it is necessary to accurately and efficiently identify turnout structural components to provide reliable data support for parameter measurements.

[0042] Figure 2 This is a specific example diagram of the dynamic measurement method for the drop value of the turnout tip rail in an embodiment of the present invention, as shown in the figure. Figure 2 As shown, feature points are calculated based on the improved PointNet++ multi-scale semantic segmentation model and the least-squares fitted turnout profile curve. The specific process is as follows: First, acquire the 3D data of the turnout; second, calculate the point cloud category based on the multi-scale semantic segmentation model to achieve automatic identification of turnout structural components; third, calculate the rail vertex by fitting the rail top curve and the track gauge point by fitting the rail side curve; finally, fit the contour curve of the neighborhood region of the track gauge point to calculate the feature points and achieve the measurement of the reduction value.

[0043] In step 102, a pre-trained multi-scale semantic segmentation model is used to classify and identify the turnout point cloud data, and output the point cloud data of the basic rail and multiple turnout structural components; the turnout structural components include wing rail, switch rail and frog rail; the multi-scale semantic segmentation model is trained based on PointNet++.

[0044] In one embodiment, the multi-scale semantic segmentation model may include a sampling layer, a grouping layer, a PointNet layer, a feature propagation layer, and an output layer;

[0045] The sampling layer is used to downsample the input turnout point cloud data using the Farthest Point Sampling (FPS) method.

[0046] The grouping layer is used to construct local area point sets from downsampled turnout point cloud data based on the K-nearest neighbor algorithm.

[0047] The PointNet layer is used to: extract features from each local region's point set;

[0048] The feature propagation layer is used to: propagate the features of a local region point set to the original points through interpolation, and output features that fuse multi-level information;

[0049] The output layer is used to classify and identify features that integrate multi-level information, and output the turnout structural components corresponding to each point.

[0050] During implementation, PointNet++ was used to perform hierarchical feature learning and point feature propagation on the turnout point cloud data.

[0051] (1) Hierarchical feature learning

[0052] Hierarchical feature learning consists of multiple Set Abstraction (SA) layers. At each layer, a set of points is processed and abstracted to generate a new set with fewer elements. The Set Abstraction layer includes a Sampling layer, a Grouping layer, and a PointNet layer. Because the density of turnout point cloud data varies significantly across different regions, the Farthest Point Sampling (FPS) strategy is employed to maintain the point cloud data structure within the partitioned region in the most balanced way possible. However, this approach disrupts the adjacency relationships of the original turnout point cloud data, thus adversely affecting the extraction of local geometric features and the learning and representation of features. Therefore, in one embodiment, the sampling layer is specifically used for:

[0053] The input turnout point cloud data is divided into multiple sub-regions; the sub-regions include the rail head region, the rail web region, and the region below the rail bottom.

[0054] The downsampling rate for each sub-region is set according to the point cloud ratio between the sub-region and the turnout point cloud data;

[0055] The FPS method is used to downsample each sub-region according to the downsampling rate of each sub-region.

[0056] This example employs a partition-based downsampling method, which includes three steps: sub-region division, sampling rate determination, and farthest point sampling. First, the turnout 3D point cloud dataset is denoted as P. Based on prior information about the rail design structure and the positions of the rail top and gauge points, the data is divided into three different sub-regions: the rail head region P... head Track waist area P waist and the area P below the track bottom base Their data volumes are |P head |、|P waist | and | Pbase Next, by proportionally reducing the amount of data in each region, we ensure that the proportions of each region in the downsampled dataset are consistent with those in the original dataset. For example, the trackhead region P head downsampling rate r head Represented as:

[0057]

[0058] Similarly, the downsampling rate r in the rail web region and the region below the rail bottom can be obtained. waist and r base .

[0059] Finally, for each region P i (Where i takes the value of head, waist, or base), downsampling is performed using the farthest point sampling method. Initialize the set of sampled points. In each sampling step, a distance S from the set of already sampled points is selected from the remaining points. i The farthest point is selected as the next sampling point. The distance formula is: for a point p∈P i and Its to the set of sampled points S i The distance is:

[0060] min{s∈S i}||ps|| (2)

[0061] Where s is the set of sampled points S i A point, S i It is the set of already sampled points. s iterates through each point in this set to calculate the distance from the point p to be sampled to the set of already sampled points S. i The minimum distance. ||·|| represents the Euclidean distance calculation, indicating the minimum distance in set S. i Find the s that minimizes "ps"; the actual intention is to iterate through the set S. i Repeat the above steps until the target amount of sampled data is reached.

[0062] The grouping layer constructs local region point sets using the K-nearest neighbor algorithm. The PointNet layer extracts features from local regions using the core network of PointNet. When the input sampling density changes, this layer can learn to combine features from regions at different scales and combine them based on the local point density.

[0063] (2) Point feature propagation

[0064] Two downsampling operations were performed in the ensemble abstraction layer. However, in semantic segmentation tasks, features from all points are required. Therefore, a hierarchical propagation strategy based on distance interpolation and cross-layer connections is adopted to propagate features from subsampled points to the original points. The feature propagation process is as follows: First, a weighted average of the inverse distances of the k nearest neighbors is calculated; then, the interpolated features of the points are concatenated with the jump connection point features of the ensemble abstraction layer to form connection features; finally, the connection features are passed through PointNet units, using a shared fully connected layer and an activation function layer to update the feature vector of each point. This process is repeated until the features are propagated to the original point set.

[0065] Figure 3 This is another specific example diagram of the dynamic measurement method for the drop value of the turnout tip rail in the embodiments of the present invention, with reference to... Figure 3 The data of the main rail, switch rail, frog rail, and wing rail are obtained through the identification results of the aforementioned turnout structural components. The reduction value is measured according to different components. First, the curve of the rail top surface is fitted based on the least squares method. Second, according to the definition of the gauge point, the gauge point located 16mm below the rail apex is calculated. Finally, based on the geometric relationship of the switch rail, the reduction value feature point is located in the neighborhood of the gauge point. The second derivative extreme point is calculated based on the curve fitted on the side of the rail head. The reduction value feature point is selected using this extreme point and then subtracted from the rail apex of the main rail or wing rail to obtain the reduction value measurement result of the switch rail.

[0066] In step 103, the rail top surface curve and rail head side surface curve are obtained by fitting the point cloud data of the basic rail and the wing rail respectively using the least squares method.

[0067] During implementation, the top surface curve and the side surface curve of the basic rail are obtained by fitting the point cloud data of the basic rail, and the top surface curve and the side surface curve of the wing rail are obtained by fitting the point cloud data of the wing rail.

[0068] In one embodiment, the track top surface curve and track head side surface curve are obtained by fitting point cloud data of the base track and wing track respectively using the least squares method, which may include:

[0069] Feature points of the rail top surface profile curve are obtained by filtering the point cloud data of the basic rail or the wing rail.

[0070] Using the standard equation of a circle, an objective function based on the least squares principle is constructed; the objective function includes the parameters to be solved.

[0071] The gradient descent method is used to iteratively adjust the parameters to be solved, so as to minimize the objective function.

[0072] Figure 4This is a schematic diagram of the rail top surface curve and rail head side curve in an embodiment of the present invention. The rail top surface profile curve is fitted using the nonlinear least squares method. Since the standard rail profile is composed of a series of circular arc curve segments, the rail vertices of the base rail and wing rail can be obtained by fitting the rail top surface tangent of the profile curve. A circular arc segment with a rail top radius of 300mm corresponding to the standard rail is selected for fitting.

[0073] Select the set of the highest N points on the contour curve of the rail top surface. Let M be the ordered set of rail vertices on the contour curve, i.e., M = {m} i =(x i ,y i Given the sequence {i = 0, 1, ..., N-1}, where N is 30 based on the sampling density. The standard equation of a circle is:

[0074] (xa) 2 +(xb) 2 =R 2 (3)

[0075] In equation (3): the center coordinates are (a, b), R is the radius of the circle, R = 300 mm. Least squares fitting curve L p The goal is to find a and b that minimize the following expression:

[0076]

[0077] We use gradient descent to iteratively adjust a and b to minimize the objective function F(a,b). The gradient is calculated as follows:

[0078]

[0079] Update parameters using the learning rate ρ: and The iteration terminates when the parameter update change is less than 0.1, and the final a and b are obtained, which are the coordinates of the center of the fitted circle.

[0080] In step 104, based on the rail top surface curve and the rail head side curve, the rail apex and gauge point are obtained, with reference to... Figure 4 The gauge point can be obtained 16mm below the top of the track.

[0081] For example, based on the profile curve L of the rail top p Calculate the horizontal tangent L0 at the top of the rail, and translate it downwards by 16mm to obtain the straight line L1. Since the rail profile P is a set of discrete points, there may be no intersection point with the straight line L1. Therefore, calculate the point P that is closest to the straight line L1. G =(x k ,y k () as the gauge point.

[0082] In step 105, based on the geometric relationship between the base rail and the switch rail, the reduction value feature point is determined based on the side curve of the rail head; the reduction value feature point reflects the vertex of the switch rail or the frog rail.

[0083] In one embodiment, based on the geometric relationship between the base rail and the switch rail, the reduction value feature points are determined based on the railhead side profile curve, including:

[0084] The extreme points of the second derivative of the railhead side profile curve are calculated; these extreme points reflect the close-fitting positions of the base rail and the switch rail.

[0085] Calculate the normal L of the curve on the side of the railhead at the extreme point. I ;

[0086] Determine the distance normal L in set S I The nearest point R; the set S is the set of base rails and switch rails in the turnout point cloud data, or the set of wing rails and center rails in the turnout point cloud data.

[0087] Starting from point R, search backwards for the point T with the largest vertical coordinate value within the gauge setting area, and record it as the decrease value feature point.

[0088] In practice, based on the structural relationship between the switch rail and the frog rail, the characteristic points for reducing the gauge value are typically located in the neighborhood of the gauge point. Therefore, an ordered set S is selected, containing point P. G and m points before and after: S={(x k-m ,y k-m ),…(x k ,y k ),…(x k+m ,y k+m )}, where x k Let represent the coordinates of the k-th point along the railway track, and let represent the coordinates of the k-th point in the vertical direction.

[0089] The calculation of the reduction value is divided into the calculation of the reduction value of the switch rail and the fixed point rail. Figure 5 This is a schematic diagram of the calculation process for the reduction value of the pointed rail in an embodiment of the present invention, as shown below. Figure 5 As shown, taking the calculation process of the switch rail reduction value as an example: First, the railhead side profile curve L is fitted using the least squares method described above. S Based on the geometric relationship of the switch rail components, the side profile of the rail head is formed by splicing the profiles of the base rail and the switch rail. Therefore, by calculating curve L... S The extreme point Q of the second derivative can be used to locate the close-fitting positions of the basic rail and the switch rail; then, the extreme point Q on the curve L is calculated. S normal L I Determine the distance normal L in set S. Ithe nearest point R, which satisfies |R-Q|<t, where t is a distance threshold set to 1. If the condition is not met, the fitting parameters need to be updated; finally, starting from point R, search backward for the point T with the maximum y value, which is the reduction value feature point, and then subtract it from the stock rail vertex to obtain the measurement result of the switch rail reduction value.

[0090] Figure 6 is a schematic diagram for calculating the reduction value feature point in the embodiment of the present invention, Figure 6 wherein the horizontal coordinate represents the abscissa of the point, and the vertical coordinate represents the ordinate of the point, showing the original data, the fitting curve, the maximum value of the second derivative, the normal direction, the point closest to the normal line, and the reduction value feature point.

[0091] In summary, the embodiments of the present invention have the following advantages:

[0092] (1) The embodiment of the present invention proposes a dynamic measurement method for the reduction value of switch and point rails of a turnout, which combines a multi-scale semantic segmentation model and a least squares fitting contour method to realize the calculation of the reduction value. This method can reduce the manual operation and maintenance cost, and provides a new technical means for railway maintenance.

[0093] (2) Based on the partition downsampling algorithm (region division → proportional sampling → FPS) with prior turnout structure, as well as the feature propagation mechanism of hierarchical feature learning and cross-layer connection, the embodiment of the present invention improves the component recognition accuracy in complex point cloud scenes.

[0094] (3) Based on the least squares rail top curve fitting algorithm, the dynamic calculation method of gauge points, and the feature point positioning strategy using the second derivative extreme point and geometric constraints, the embodiment of the present invention realizes the reduction value measurement with an accuracy of ±0.3mm, which meets the millimeter-level detection standard for high-speed railways.

[0095] The embodiment of the present invention further provides a dynamic measurement device for the reduction value of switch and point rails of a turnout, as described in the following embodiments. Since the principle of the device to solve the problem is similar to that of the dynamic measurement method for the reduction value of switch and point rails of a turnout, the implementation of the device can refer to the implementation of the dynamic measurement method for the reduction value of switch and point rails of a turnout, and repeated descriptions will not be omitted.

[0096] Figure 7 is a schematic diagram of the dynamic measurement device for the reduction value of switch and point rails of a turnout in the embodiment of the present invention, as shown in Figure 7 , the device 700 includes:

[0097] a data acquisition module 701, configured to acquire turnout point cloud data;

[0098] The segmentation model processing module 702 is used to classify and identify turnout point cloud data using a pre-trained multi-scale semantic segmentation model, and output point cloud data of the basic rail and multiple turnout structural components; the turnout structural components include wing rail, switch rail and frog rail; the multi-scale semantic segmentation model is trained based on PointNet++.

[0099] The reduction value calculation module 703 is used to obtain the rail top surface curve and rail head side curve by fitting the point cloud data of the base rail and wing rail respectively based on the least squares method; obtain the rail vertex and gauge point based on the rail top surface curve and rail head side curve; determine the reduction value feature point based on the rail head side curve according to the geometric relationship between the base rail and the switch rail; the reduction value feature point reflects the vertex of the switch rail or the frog rail; and determine the reduction value of the switch rail and the reduction value of the fixed frog rail by using the height difference between the rail vertex and the reduction value feature point.

[0100] In one embodiment, the multi-scale semantic segmentation model includes a sampling layer, a grouping layer, a PointNet layer, a feature propagation layer, and an output layer;

[0101] The sampling layer is used to downsample the input turnout point cloud data using the FPS method;

[0102] The grouping layer is used to construct local area point sets from downsampled turnout point cloud data based on the K-nearest neighbor algorithm.

[0103] The PointNet layer is used to: extract features from each local region's point set;

[0104] The feature propagation layer is used to: propagate the features of a local region point set to the original points through interpolation, and output features that fuse multi-level information;

[0105] The output layer is used to classify and identify features that integrate multi-level information, and output the turnout structural components corresponding to each point.

[0106] In one embodiment, the sampling layer is specifically used for:

[0107] The input turnout point cloud data is divided into multiple sub-regions; the sub-regions include the rail head region, the rail web region, and the region below the rail bottom.

[0108] The downsampling rate for each sub-region is set according to the point cloud ratio between the sub-region and the turnout point cloud data;

[0109] The FPS method is used to downsample each sub-region according to the downsampling rate of each sub-region.

[0110] In one embodiment, the reduction value calculation module 703 is specifically used for:

[0111] Feature points of the rail top surface profile curve are obtained by filtering the point cloud data of the basic rail or the wing rail.

[0112] Using the standard equation of a circle, an objective function based on the least squares principle is constructed; the objective function includes the parameters to be solved.

[0113] The gradient descent method is used to iteratively adjust the parameters to be solved, so as to minimize the objective function.

[0114] In one embodiment, the reduction value calculation module 703 is specifically used for:

[0115] The extreme points of the second derivative of the railhead side profile curve are calculated; these extreme points reflect the close-fitting positions of the base rail and the switch rail.

[0116] Calculate the normal L of the curve on the side of the railhead at the extreme point. I ;

[0117] Determine the distance normal L in set S I The nearest point R; the set S is the set of base rails and switch rails in the turnout point cloud data, or the set of wing rails and center rails in the turnout point cloud data.

[0118] Starting from point R, search backwards for the point T with the largest vertical coordinate value within the gauge setting area, and record it as the decrease value feature point.

[0119] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned dynamic measurement method for the drop value of the turnout tip rail.

[0120] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for dynamically measuring the drop value of the turnout tip rail.

[0121] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-mentioned dynamic measurement method for the drop value of the turnout tip rail.

[0122] In this embodiment of the invention, the PointNet++ model is first used to classify and identify the point cloud data of the turnout, outputting point cloud data of multiple turnout structural components, thus improving the identification accuracy of each component in the turnout point rail scenario. Then, an adaptive least squares fitting algorithm is used to extract the rail vertex, gauge point, and drop value feature points, achieving high-precision measurement of the point rail drop value and the fixed point rail drop value. Experimental verification shows that the method achieves a measurement accuracy of ±0.3mm under dynamic detection conditions, meeting the millimeter-level detection standard for high-speed railway turnouts, and improving the accuracy and efficiency of turnout point rail drop value measurement.

[0123] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0124] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0125] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0126] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0127] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic measurement method for the drop value of the turnout tip rail, characterized in that, include: Acquire turnout point cloud data; A pre-trained multi-scale semantic segmentation model is used to classify and identify turnout point cloud data, outputting point cloud data of the base rail and multiple turnout structural components; the turnout structural components include wing rail, switch rail, and frog rail; the multi-scale semantic segmentation model is trained based on PointNet++. Based on the least squares method, the rail top surface curve and rail head side surface curve are obtained by fitting the point cloud data of the base rail and the wing rail respectively. Based on the curves of the top surface of the rail and the side surface of the rail head, the rail apex and gauge point are obtained. Based on the geometric relationship between the base rail and the switch rail, the characteristic points of the reduction value are determined based on the side curve of the rail head. The decreasing value characteristic point reflects the apex of the switch rail or the center rail; The drop value of the switch rail is determined by using the height difference between the rail apex and the drop value feature point, and the drop value of the center rail is fixed. Specifically, based on the geometric relationship between the base rail and the switch rail, the characteristic points for the reduction value are determined based on the side curve of the railhead, including: The extreme points of the second derivative of the railhead side profile curve are calculated; these extreme points reflect the close-fitting positions of the base rail and the switch rail. Calculate the normal to the curve on the side of the railhead at the extreme point. ; Determine the distance normal in set S The nearest point R; the set S is the set of base rails and switch rails in the turnout point cloud data, or the set of wing rails and center rails in the turnout point cloud data. Starting from point R, search backwards for the point T with the largest vertical coordinate value within the gauge point setting area, and record it as the decrease value feature point.

2. The method as described in claim 1, characterized in that, The multi-scale semantic segmentation model includes a sampling layer, a grouping layer, a PointNet layer, a feature propagation layer, and an output layer; The sampling layer is used to downsample the input turnout point cloud data using the farthest point sampling (FPS) method. The grouping layer is used to construct local area point sets from downsampled turnout point cloud data based on the K-nearest neighbor algorithm. The PointNet layer is used to: extract features from each local region's point set; The feature propagation layer is used to: propagate the features of a local region point set to the original points through interpolation, and output features that fuse multi-level information; The output layer is used to classify and identify features that integrate multi-level information, and output the turnout structural components corresponding to each point.

3. The method as described in claim 2, characterized in that, The sampling layer is specifically used for: The input turnout point cloud data is divided into multiple sub-regions; the sub-regions include the rail head region, the rail web region, and the region below the rail bottom. The downsampling rate for each sub-region is set according to the point cloud ratio between the sub-region and the turnout point cloud data; The FPS method is used to downsample each sub-region according to the downsampling rate of each sub-region.

4. The method as described in claim 1, characterized in that, Based on the least squares method, the rail top surface curve and rail head side surface curve are obtained by fitting point cloud data of the base rail and wing rail respectively, including: Feature points of the rail top surface profile curve are obtained by filtering the point cloud data of the basic rail or the wing rail. Using the standard equation of a circle, an objective function based on the least squares principle is constructed; the objective function includes the parameters to be solved. The gradient descent method is used to iteratively adjust the parameters to be solved, so as to minimize the objective function.

5. A dynamic measuring device for the drop value of the turnout tip rail, characterized in that, include: The data acquisition module is used to acquire turnout point cloud data; The segmentation model processing module is used to classify and identify turnout point cloud data using a pre-trained multi-scale semantic segmentation model, and output point cloud data of the base rail and multiple turnout structural components; the turnout structural components include wing rail, switch rail and frog rail; the multi-scale semantic segmentation model is trained based on PointNet++. The reduction value calculation module is used to obtain the rail top surface curve and rail head side surface curve by fitting the point cloud data of the base rail and wing rail respectively based on the least squares method. Based on the top surface curve and the side surface curve of the rail head, the rail apex and gauge point are obtained; based on the geometric relationship between the basic rail and the switch rail, the reduction value characteristic points are determined based on the side surface curve of the rail head. The drop value feature point reflects the apex of the switch rail or the frog rail; the drop value of the switch rail is determined and the drop value of the frog rail is fixed by using the height difference between the rail apex and the drop value feature point. Specifically, the reduction value calculation module is used for: The extreme points of the second derivative of the railhead side profile curve are calculated; these extreme points reflect the close-fitting positions of the base rail and the switch rail. Calculate the normal to the curve on the side of the railhead at the extreme point. ; Determine the distance normal in set S The nearest point R; the set S is the set of base rails and switch rails in the turnout point cloud data, or the set of wing rails and center rails in the turnout point cloud data. Starting from point R, search backwards for the point T with the largest vertical coordinate value within the gauge point setting area, and record it as the decrease value feature point.

6. The apparatus as claimed in claim 5, characterized in that, The multi-scale semantic segmentation model includes a sampling layer, a grouping layer, a PointNet layer, a feature propagation layer, and an output layer; The sampling layer is used to downsample the input turnout point cloud data using the FPS method; The grouping layer is used to construct local area point sets from downsampled turnout point cloud data based on the K-nearest neighbor algorithm. The PointNet layer is used to: extract features from each local region's point set; The feature propagation layer is used to: propagate the features of a local region point set to the original points through interpolation, and output features that fuse multi-level information; The output layer is used to classify and identify features that integrate multi-level information, and output the turnout structural components corresponding to each point.

7. The apparatus as claimed in claim 6, characterized in that, The sampling layer is specifically used for: The input turnout point cloud data is divided into multiple sub-regions; the sub-regions include the rail head region, the rail web region, and the region below the rail bottom. The downsampling rate for each sub-region is set according to the point cloud ratio between the sub-region and the turnout point cloud data; The FPS method is used to downsample each sub-region according to the downsampling rate of each sub-region.

8. The apparatus as claimed in claim 5, characterized in that, The reduction value calculation module is specifically used for: Feature points of the rail top surface profile curve are obtained by filtering the point cloud data of the basic rail or the wing rail. Using the standard equation of a circle, an objective function based on the least squares principle is constructed; the objective function includes the parameters to be solved. The gradient descent method is used to iteratively adjust the parameters to be solved, so as to minimize the objective function.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 4.

11. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 4.

Images