Heavy-load ballast track bed elasticity detection method and system based on laser measurement

Abstract

The invention relates to the technical field of ballast track bed elasticity detection, in particular to a heavy-load ballast track bed elasticity detection method and system based on laser measurement, and the method comprises the steps: S1, collecting the amplitude data of a train and a track in a running state; s2, according to the amplitude data and the operation parameters of the train, performing simulation operation to obtain an elevation reference value of the track; s3, performing three-dimensional scanning on the rail by taking the elevation reference value as a measurement zero point in the height direction of the rail to generate continuous dynamic point cloud data of elastic deformation of the rail; and S4, outputting an elastic deformation value of the track from the point cloud data by using a deep learning algorithm model, and representing the elastic deformation value by an oscillogram to realize rapid detection of the elastic deformation of the track. According to the invention, the problem that the elasticity of the heavy-load ballast track bed cannot be dynamically, efficiently and accurately detected in the prior art can be effectively solved.

Description

Technical Field

[0001] The invention relates to the technical field of ballasted track bed elasticity detection, and in particular to a heavy-load ballasted track bed elasticity detection method and system based on laser measurement. Background Art

[0003] Railways are affected by complex geological and topographical conditions along the line, natural disasters, and many unfavorable factors in their design, construction, and operation management.

[0004] Therefore, it is urgent to develop a method for rapid detection of track elastic deformation suitable for heavy-load freight railways to provide scientific and accurate data support for railway maintenance. Summary of the invention

[0005] Technical issues solved

[0006] In view of the above-mentioned shortcomings of the prior art, the present invention provides a method and system for detecting the elasticity of a heavy-loaded ballasted track bed based on laser measurement, which can effectively solve the problem that the prior art means cannot dynamically, efficiently and accurately detect the elasticity of a heavy-loaded ballasted track bed.

[0007] Technical Solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] On one hand, the present invention provides a method for detecting elasticity of a heavy-loaded ballast track bed based on laser measurement, comprising step S1: collecting amplitude data of a train and a track in a running state;

[0010] Step S2: According to the amplitude data and the running parameters of the train, a simulation operation is performed to obtain a reference value of the elevation of the track;

[0011] Step S3: performing a three-dimensional scan of the track using the elevation reference value as the measurement zero point in the height direction of the track to generate continuous dynamic point cloud data of the elastic deformation of the track;

[0012] Step S4: Apply the deep learning algorithm model to the point cloud data to output the elastic deformation value of the track and represent it in a waveform graph to achieve rapid detection of the elastic deformation of the track.

[0013] Another aspect of the present invention is a laser-measured heavy-duty ballast track bed elasticity detection system, the detection system comprising a collection module, including a laser rangefinder for collecting amplitude data of trains and tracks, and a line scanning laser profiler for three-dimensionally scanning the tracks;

[0014] and a data processing module, the data processing module being used to pre-process the collected amplitude data and point cloud data;

[0015] and a model building module, for building a model of a track and a train according to the amplitude data and the running parameters of the train, and running the deep learning model;

[0016] and a data calculation module, for simulating and calculating the elevation reference value of the track, and for executing a deep learning algorithm to output the elastic deformation value of the track;

[0017] And a graphics drawing module is used to draw the amplitude waveform line of the track and draw the elastic deformation waveform diagram of the track.

[0018] Furthermore, in step S1, the amplitude of the train includes the amplitude of the wheels and the carriage of the train, and corresponding vibration data are obtained by independently installing sensors on the wheels and the carriage.

[0019] Further, in step S2, the accurate elevation reference value is obtained by the following steps:

[0020] S21: The acquired amplitude data needs to be cleaned and aligned;

[0021] S22: analyzing the variation trend of the amplitude according to different road conditions of the track bed, and drawing an amplitude waveform line;

[0022] S23: Building a model based on the running speed of the train and other parameters, and performing simulation calculation to obtain a reference value of the track elevation.

[0023] Furthermore, in step S3, a rail elastic deformation database may be established, wherein the database is used to collect and analyze continuous dynamic point cloud data of the rail elastic deformation and store analysis and detection results.

[0024] Further, in step S4, the elastic deformation value is obtained by the following steps:

[0025] S41: performing noise removal and outlier removal processing on the point cloud data;

[0026] S42: Use geometric methods to extract key feature points in the point cloud and restore the three-dimensional geometric structure and surface features of the object from the scattered point cloud data;

[0027] S43: A deep learning model based on a convolutional neural network is used to process point cloud data to obtain typical response characteristics of track deformation. A large amount of data in the track elastic deformation database is used for training to build an elastic deformation analysis system. The analysis system improves the network architecture's ability to perceive elastic deformation information and filter background clutter, thereby achieving intelligent recognition of track elastic deformation and rapid dynamic waveform output.

[0028] Furthermore, in the process of processing the point cloud data, a bilateral filtering algorithm is used, and spatial information and grayscale similarity are considered at the same time to achieve the purpose of edge-preserving denoising, wherein the bilateral filtering algorithm targets the spatial domain and pixel range domain of the image.

[0029] Furthermore, a plurality of the line scanning laser profilers are arranged on the train, and the track is quickly scanned by a line scanning laser light source, wherein the elevation reference value of the track is used as a measurement and calculation zero point and is introduced into the line scanning laser profiler.

[0030] Beneficial Effects

[0031] Compared with the known public technology, the technical solution provided by the present invention has the following beneficial effects:

[0032] This solution uses a laser rangefinder to collect the amplitude of the train and track, and uses a line-scan laser profiler to perform a three-dimensional scan of the track. By obtaining continuous dynamic point cloud data of the track, the elastic deformation value of the track is output through a deep learning algorithm model, thereby realizing intelligent recognition of the elastic deformation of the track and rapid dynamic waveform output. It provides efficient, scientific and accurate data support for railway maintenance in terms of elastic detection of heavy-loaded ballasted trackbeds. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 It is a schematic diagram of the flow of the rail elastic deformation detection method of the present invention;

[0035] Figure 2 It is a schematic diagram of a rail elasticity detection system of the present invention;

[0036] Figure 3 It is a schematic diagram of the scanning track of the line scanning laser profile detector of the present invention;

[0037] Figure 4 This is a schematic diagram of the arrangement of the line scanning laser profile detector of the present invention on a train;

[0038] The numbers in the figure represent: 10, carriage; 20, wheel; 30, track; 40, line scanning laser profile detector. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] The present invention will be further described below in conjunction with the embodiments.

[0041] Example:

[0042] The present invention provides a method and system for detecting the elasticity of a heavy-loaded ballasted track bed based on laser measurement. The method and system of the present invention can be used for the rapid detection of the elasticity of ballasted track beds under different road conditions. The detection system is obtained by training based on a deep learning model. By inputting the collected point cloud data of a train and a track 30, the elastic deformation value of the track 30 can be quickly output and expressed as a waveform diagram, thereby realizing the rapid detection of the elastic deformation of the track 30. The existing technical means are effectively used to perform periodic maintenance with low precision, poor pertinence, and a large amount of waste of manpower and material resources.

[0043] Reference Figure 1-4 The steps of the ballasted track bed elasticity detection method of the present invention include:

[0044] Step S1: collecting the amplitude data of the train and the track 30 in the running state, wherein the amplitude of the train includes the amplitude of the wheels 20 and the carriage 10 of the train, and obtaining the corresponding vibration data by independently installing sensors on the wheels 20 and the carriage 10.

[0045] Step S2: According to the amplitude data and the train operation parameters, the elevation reference value of the track 30 is obtained by simulation calculation; wherein, the elevation reference value is a comprehensive simulation estimation parameter, and multiple simulations and adjustments are required to obtain the optimal solution. It is necessary to consider as many as possible: the train operation parameters: including the weight, speed, acceleration, etc. of the train, the amplitude data of the track 30, and the physical properties of the track 30: such as stiffness, damping coefficient, etc.

[0046] Step S3: Take the elevation reference value as the measurement zero point in the height direction of the track 30 to perform a three-dimensional scan of the track 30, and generate continuous dynamic point cloud data of the elastic deformation of the track 30. A database of elastic deformation of the track 30 can be established for a large amount of point cloud data. The database is used to collect and analyze the continuous dynamic point cloud data of elastic deformation of the track 30, and store the analysis and detection results; and provide data support for subsequent deep learning model training.

[0047] Step S4: Apply the deep learning algorithm model to the point cloud data to output the elastic deformation value of the track 30 and represent it in a waveform diagram to achieve rapid detection of the elastic deformation of the track 30. The deep learning algorithm model is obtained through training of a large amount of point cloud data.

[0048] The present invention provides a system for performing a laser measurement method for elastic detection of a heavy-duty ballasted track bed. The system mainly includes an acquisition module, including a laser rangefinder for acquiring amplitude data of a train and a track 30 , and a line scanning laser profiler for performing a three-dimensional scanning of the track 30 .

[0049] and a data processing module, the data processing module is used to pre-process the collected amplitude data and point cloud data;

[0050] and a model building module for building models of the track 30 and the train according to the amplitude data and the operating parameters of the train, and running the deep learning model;

[0051] and a data calculation module, for simulating and calculating the elevation reference value of the track 30 , and for executing a deep learning algorithm to output the elastic deformation value of the track 30 ;

[0052] And a graphics drawing module, using MATLAB, Python or other drawing software, is used to draw the amplitude waveform line of the track 30 and the elastic deformation waveform diagram of the track 30.

[0053] In measuring the amplitude, a laser rangefinder is installed on the carriage 10, the wheel 20 or the track 30, and a high-precision reflector (such as a prism or a reflective plate) is installed at a fixed position on or near the track 30. When the train passes, the laser rangefinder records the distance change between the reflector and the rangefinder. This distance change can indirectly reflect the vibration of the track 30.

[0054] In this embodiment, preferably 7-12 line scanning laser profilers are evenly arranged on the train, and the track 30 is quickly scanned by the line scanning laser light source, wherein the elevation reference value of the track 30 is used as the measurement and calculation zero point and is imported into the line scanning laser profiler. The line scanning laser profiler projects a laser plane onto the surface of the track 30, and uses a CMOS camera to receive the changes in its reflected light, so as to measure the height, height difference, width and other contours (cross-sectional shapes) in a non-contact manner. Through the relative movement between the track 30 and the profiler, the continuously acquired contour data is image processed and reconstructed to obtain the three-dimensional point cloud data of the track 30, thereby achieving high-precision measurement and detection.

[0055] The following is the specific implementation method of each step in the detection method of this scheme.

[0056] In step S2, the precise elevation reference value is obtained by the following steps:

[0057] S21: The acquired amplitude data needs to be cleaned and aligned;

[0058] S22: Analyze the variation trend of the amplitude according to the different road conditions of the track 30, and draw the amplitude waveform line; wherein, according to the different road conditions of the road conditions (such as good, general, poor, etc.), classify the amplitude data, and pay attention to the changes of the amplitude, frequency, duration and other parameters.

[0059] S23: A model is established in combination with the running speed of the train and other parameters, and a simulation operation is performed to obtain the elevation reference value of the track 30. Other parameters include the traction force, braking force, load condition, etc. of the train, which may also affect the vibration characteristics of the train, and therefore should also be collected when conditions permit.

[0060] In step S3, according to the actual situation of the railway track 30 on site, this embodiment preferably adopts a special line scanning laser light source suitable for outdoor railway application scenarios, temperature, waterproof level, speed, vibration, light interference, dust prevention, etc. At the same time, the number of points arranged by the line scanning laser profile detector 40 is 7-12, distributed on the locomotive from front to back. By quickly scanning the track 30 with the line scanning laser profile detector 40, continuous dynamic point cloud data of the elastic deformation of the track 30 can be generated, the data is collected and analyzed, and a railway track 30 elastic deformation database is established for storing and analyzing the detection results.

[0061] In addition, in step S4, the elastic deformation value is obtained by the following steps:

[0062] S41: De-noising and removing outliers on the point cloud data;

[0063] S42: Use geometric methods to extract key feature points in the point cloud and restore the three-dimensional geometric structure and surface features of the object from the scattered point cloud data;

[0064] S43: A deep learning model based on a convolutional neural network is used to process point cloud data to obtain typical response characteristics of the deformation of the track 30. A large amount of data in the elastic deformation database of the track 30 is used for training to build an elastic deformation analysis system. The analysis system improves the network architecture's ability to perceive elastic deformation information and filter background clutter, thereby achieving intelligent recognition of the elastic deformation of the track 30 and rapid dynamic waveform output.

[0065] Specifically, in the process of processing point cloud data, a bilateral filtering algorithm is used, which is a compromise processing method that combines the spatial proximity of the image and the similarity of pixel values, while considering the spatial information and grayscale similarity to achieve the purpose of edge-preserving denoising. Among them, the bilateral filtering algorithm targets the spatial domain and pixel range domain of the image.

[0066] When designing the algorithm, there will be two weights, Ws and Wr, and the algorithm formula is as follows:

[0067]

[0068] In the formula, p is the current point. The reason why it is a vector is that sometimes the image is not only a grayscale image, but also a three-color image. Therefore, a vector is used to represent the multi-dimensional space of p. I is the image, Ip represents the point p of the image I, S is the neighborhood set of p, q is a point in the neighborhood, and BF is the output image.

[0069] In step S42, geometric methods are used to extract key feature points in the point cloud, such as edges, corners, etc. These feature points are crucial for subsequent surface reconstruction because they represent important geometric information on the surface of the track 30; the point cloud data is further processed by image processing techniques (such as edge detection, texture analysis, etc.) to improve the accuracy and representativeness of feature extraction. The scattered point cloud data is then segmented according to surface features to form multiple point cloud subsets with similar features. This helps the subsequent surface reconstruction process because each subset can be independently fitted to the surface; generating a smooth and continuous surface model.

[0070] Curvature is a measure of the flatness of the surface of the point cloud model, reflecting the concave and convex state of the point cloud surface. In this embodiment, the method of fitting the k-neighborhood K(pi) surface with a quadratic surface is used to estimate the curvature. The estimation of curvature information mainly includes the principal curvatures K1 and K2, the average curvature H and the Gaussian curvature K. In general, if only the average curvature is required, the average can be estimated by principal component analysis.

[0071] The calculation formula of average curvature H is as follows:

[0072]

[0073] This solution uses a laser rangefinder to collect the amplitude of the train and the track 30, and uses a line scanning laser profiler to perform a three-dimensional scan of the track 30. By obtaining continuous dynamic point cloud data of the track 30, the elastic deformation value of the track 30 is output through a deep learning algorithm model, thereby realizing intelligent recognition of the elastic deformation of the track 30 and rapid dynamic waveform output. It is aimed at the elastic detection of heavy-loaded ballasted trackbeds.

[0074] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention 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. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting elasticity of heavy-loaded ballast track bed based on laser measurement, characterized in that: include: Step S1: collecting amplitude data of the train and track in operation; Step S2: According to the amplitude data and the running parameters of the train, a simulation operation is performed to obtain a reference value of the elevation of the track; Step S3: performing a three-dimensional scan of the track using the elevation reference value as the measurement zero point in the height direction of the track to generate continuous dynamic point cloud data of the elastic deformation of the track; Step S4: Apply the deep learning algorithm model to the point cloud data to output the elastic deformation value of the track and represent it in a waveform graph to achieve rapid detection of the elastic deformation of the track.

2. The method for detecting elasticity of heavy-loaded ballast track bed based on laser measurement according to claim 1 is characterized in that: In step S1, the amplitude of the train includes the amplitude of the wheels and the carriage of the train, and corresponding vibration data are obtained by independently installing sensors on the wheels and the carriage.

3. The method for detecting elasticity of heavy-loaded ballast track bed based on laser measurement according to claim 1 is characterized in that: In step S2, the accurate elevation reference value is obtained by the following steps: S21: The acquired amplitude data needs to be cleaned and aligned; S22: analyzing the variation trend of the amplitude according to different road conditions of the track bed, and drawing an amplitude waveform line; S23: Building a model based on the running speed of the train and other parameters, and performing simulation calculation to obtain a reference value of the track elevation.

4. The method for detecting elasticity of heavy-loaded ballast track bed based on laser measurement according to claim 1 is characterized in that: In step S3, a rail elastic deformation database may be established, wherein the database is used to collect and analyze continuous dynamic point cloud data of the rail elastic deformation and store analysis and detection results.

5. The method for detecting elasticity of heavy-loaded ballast track bed based on laser measurement according to claim 1 is characterized in that: In step S4, the elastic deformation value is obtained by the following steps: S41: performing noise removal and outlier removal processing on the point cloud data; S42: Use geometric methods to extract key feature points in the point cloud and restore the three-dimensional geometric structure and surface features of the object from the scattered point cloud data; S43: A deep learning model based on a convolutional neural network is used to process point cloud data to obtain typical response characteristics of track deformation. A large amount of data in the track elastic deformation database is used for training to build an elastic deformation analysis system. The analysis system improves the network architecture's ability to perceive elastic deformation information and filter background clutter, thereby achieving intelligent recognition of track elastic deformation and rapid dynamic waveform output.

6. The method for detecting elasticity of heavy-loaded ballast track bed based on laser measurement according to claim 5 is characterized in that: In the process of processing the point cloud data, a bilateral filtering algorithm is used, and spatial information and grayscale similarity are considered at the same time to achieve the purpose of edge-preserving denoising, wherein the bilateral filtering algorithm targets the spatial domain and pixel range domain of the image.

7. A laser measurement system for detecting elasticity of heavy-loaded ballast track bed, characterized in that: The detection system comprises: A collection module, including a laser rangefinder for collecting amplitude data of the train and the track, and a line scanning laser profiler for three-dimensional scanning of the track; A data processing module, wherein the data processing module is used to pre-process the collected amplitude data and point cloud data; A model building module, used to build a model of the track and the train according to the amplitude data and the running parameters of the train, and to run the deep learning model; A data calculation module, used for simulating and calculating the elevation reference value of the track, and for executing a deep learning algorithm to output the elastic deformation value of the track; The graphics drawing module is used to draw the amplitude waveform line of the track and the elastic deformation waveform diagram of the track.

8. The laser measurement system for detecting elasticity of heavy-loaded ballast track bed according to claim 7, characterized in that: A plurality of line scanning laser profilers are arranged on the train, and the track is quickly scanned by a line scanning laser light source, wherein the elevation reference value of the track is used as a measurement and calculation zero point and is introduced into the line scanning laser profiler.

Images