A method, system, storage medium, and electronic equipment for testing rail fasteners.

Abstract

This invention provides a method, system, storage medium, and electronic device for detecting rail fasteners. The method includes acquiring a three-dimensional point cloud map of the track; using the highest point at each location in the three-dimensional point cloud map as a reference plane to calculate the corresponding ground plane; identifying the point cloud representing the rail to obtain a rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other; identifying interfering point cloud blocks representing non-rail bodies; scanning the track using a track inspection instrument to obtain a chord data map; based on the chord data map, identifying the measurement point cloud blocks representing fasteners; calculating the geometric positional relationship between each data point in the measurement point cloud block relative to the reference plane and the ground plane, as well as the geometric positional relationship between adjacent data points, to determine the type of fastener; periodically monitoring the geometric positional relationship between each data point in the measurement point cloud block relative to the reference plane and the ground plane, as well as the geometric positional relationship between adjacent data points, to detect the integrity of each fastener.

Description

Technical Field

[0001] This invention relates to the field of rail inspection technology, and in particular to a method, system, storage medium, and electronic device for inspecting rail fasteners. Background Technology

[0002] The rail transit industry has developed rapidly in recent years. The operating speed of rail transit systems has increased, the construction mileage has increased, and the lines have become increasingly busy. This has led to increased pressure on the tracks. Tracks are generally made up of multiple sections of steel rails connected in sequence. The junctions of two steel rails are connected by fasteners. Therefore, the condition of the fasteners needs to be checked frequently to ensure the safety of the tracks.

[0003] In existing technologies, defects in fasteners are generally detected using image processing, a visual approach that requires a large amount of data input to train the defect recognition model. This results in a large training load for the model, and the detection accuracy is greatly affected by ambient light and the quality of the model, leading to various errors during the detection process. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method for detecting rail fasteners. This method addresses the problem that existing technologies typically employ image processing as a visual approach to detect fastener defects. This requires extensive data input to train a defect recognition model, resulting in a large training load and significant impact on detection accuracy due to ambient light and model quality. Consequently, various errors occur during the detection process.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0006] A method for inspecting rail fasteners includes the following steps:

[0007] The track is scanned sequentially using a lidar to obtain a three-dimensional point cloud map of the track.

[0008] The highest point at each location in the three-dimensional point cloud map is taken as the reference plane to calculate the corresponding ground plane;

[0009] Feature analysis is performed on the three-dimensional point cloud map to identify the point cloud representing the rail, thus obtaining a rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other;

[0010] The vertical and horizontal positions of the rail point cloud blocks are analyzed to identify interfering point cloud blocks that do not represent the rail body itself.

[0011] The track is scanned using a track inspection instrument to obtain a chord data map. Based on the chord data map, the interference point cloud blocks representing the fasteners are identified and used as the measurement point cloud blocks.

[0012] The geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, are calculated to determine the type of fastener;

[0013] The geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, are monitored periodically to detect the integrity of each fastener.

[0014] According to one aspect of the above technical solution, the step of sequentially scanning the track with a lidar to obtain a three-dimensional point cloud map of the track further includes:

[0015] The 3D point cloud image is denoised using direct-pass filtering and voxel filtering to reduce the number of data points in the 3D point cloud image.

[0016] According to one aspect of the above technical solution, the step of taking the highest point at each location of the three-dimensional point cloud map as a reference plane to calculate the corresponding ground plane specifically includes:

[0017] At the highest point of the three-dimensional point cloud map, three points are randomly selected to construct multiple reference planes. Each data point on the reference plane is extracted as the top surface data point, and all data points in the vertical direction of the top surface data point are taken as undetermined data points.

[0018] Set a first height threshold, and calculate the first vertical height from each of the top surface data points to each of the undetermined data points in its vertical direction;

[0019] Calculate whether the first vertical height is greater than the first height threshold. If so, then take the undetermined data point as a ground data point.

[0020] An optimization formula is introduced to calculate whether the number of ground data points meets the predetermined number based on the current environment of the orbit.

[0021] If the predicted probability is less than the preset probability, then three points are reselected to construct the reference plane, and the number of ground data points is iteratively calculated until the predicted probability is greater than the preset probability.

[0022]

[0023] Where P represents the prediction probability, γ represents the number of predetermined ground data points, θ represents the number of ground data points that are exceeded or missing, q represents the number of all undetermined data points in the vertical direction, and Q represents the number of iterations.

[0024] The plane formed by the highest ground data points obtained in the last calculation is taken as the ground surface plane.

[0025] According to one aspect of the above technical solution, the feature analysis of the three-dimensional point cloud map is performed to find the point cloud representing the rail location, thereby obtaining a rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other, specifically including:

[0026] Using the ground surface as a reference plane and the extension direction of the rails in the track as the X-axis, multiple starting points x in the X-axis are proposed. n And set a sliding window (x n ,0);

[0027] Based on the preset width of the rail as the search command, the sliding window is slid along the Y-axis of the rail width direction to obtain multiple rail bottom window points;

[0028] (w 11 w 22 w 33 ......w nm )

[0029] Among them, w nm This indicates starting from the nth starting point x. n The m-th bottom window pane point of the starting rail;

[0030] All data points above the window panes on the bottom surface of the rail are extracted and regarded as rail body data points to obtain a rail point cloud map. Then, two rail point cloud blocks are extracted from the rail point cloud map.

[0031] According to one aspect of the above technical solution, the analysis of the vertical and horizontal positions of the rail point cloud blocks to identify interfering point cloud blocks that do not represent the rail body specifically includes:

[0032] Calculate the second vertical height of each data point of the rail point cloud block with its left and right adjacent points;

[0033] H ij1 =|z ij -z i(j+1) |

[0034] H ij2 =|z ij -z i(j-1) |;

[0035] Among them, Z ij H represents the height of the data point at position i on the X-axis and position j on the Y-axis. ij1 H is the second vertical height between the data point and its right neighbor. ij2 The second vertical height between the data point and its left adjacent point;

[0036] A second height threshold is set. If the second vertical height is within the range of the second height threshold, then the left adjacent point or the right adjacent point is regarded as a ballast point cloud block to be determined.

[0037] Import historical data from the undetermined ballast point cloud block and perform a reliability assessment on the undetermined ballast point cloud block;

[0038]

[0039] Where t is time, k is the number of measurements, and Z is the time interval. tij Let be the height of the data point at position ij at time t. To measure the average height of data points at position ij within a time range, δ ij Let be the standard deviation of the height of the data point at position ij;

[0040] If the data points in the undetermined ballast point cloud block meet the credibility judgment criteria, then the undetermined ballast point cloud block is regarded as a ballast point cloud block, and the ballast point cloud block is excluded from the rail point cloud block.

[0041]

[0042] Where μ is the error range, 2≤μ≤3;

[0043] In the updated rail point cloud block, the horizontal distance between the edge data points at each horizontal height is calculated. Referring to the rail design standard, edge data points whose horizontal distance is greater than the rail design standard are extracted and regarded as undetermined foreign object point cloud blocks.

[0044] Import historical data of the point cloud block of the undetermined foreign object, and evaluate the credibility of the point cloud block of the undetermined foreign object;

[0045]

[0046] Among them, X tij Let be the vertical horizontal position of the data point at position ij at time t. To measure the average vertical horizontal position of data points at position ij within the time range, Δ ij Let be the standard deviation of the vertical horizontal position of the data point at position ij;

[0047] If the data points in the undetermined foreign object point cloud block meet the credibility judgment criteria, then the undetermined foreign object point cloud block is regarded as a foreign object point cloud block.

[0048]

[0049] The ballast point cloud block and the foreign object point cloud block are considered together as interference point cloud blocks.

[0050] According to one aspect of the above technical solution, the method of using a track inspection instrument to scan the track to obtain a chord data map, and based on the chord data map, identifying interference point cloud blocks representing fasteners and using them as measurement point cloud blocks, specifically includes:

[0051] The track is scanned using a track inspection instrument to obtain a chord data map, and a chord data matrix T is constructed based on the chord data map;

[0052]

[0053] Where N represents the total number of scans by the track inspection instrument. For scanning and sampling data;

[0054] Based on the matrix binary recursive construction method, the chord data matrix is ​​decomposed by singular value decomposition to obtain similar singular values ​​and key singular values, and then the similar signals ε and key signals λ corresponding to the similar singular values ​​and key singular values ​​are obtained respectively.

[0055] The similar signals are decomposed step by step to obtain the similar signals ε at each level. a and the key signal λ a Where a is the number of factorizations;

[0056] Construct a key signal matrix D based on the key signals at all levels. a ;

[0057] In the key signal matrix D a Search for signals with abrupt peak changes and consider these signals as the first undetermined signals.

[0058] Analyze signals with the same peaks and consistent spacing in the first undetermined signal, and regard them as the second undetermined signal;

[0059] Referring to the fastener design specifications, the second undetermined signal whose spacing conforms to the fastener design specifications is regarded as the fastener signal, and the interference point cloud block at the position of the fastener signal is regarded as the measurement point cloud block.

[0060] According to one aspect of the above technical solution, calculating the geometric positional relationship between each data point in the measurement point cloud block relative to the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, to determine the type of the fastener, specifically includes:

[0061] Calculate the vertical distance between each data point in the measurement point cloud block and the reference plane and the ground surface plane respectively. Referring to the fastener design specification, data points whose vertical distance conforms to the fastener design specification are regarded as data points to be verified.

[0062] Hij3 =z ij

[0063] H ij4 =|z ij -z′ ij |;

[0064] Where, z′ ij H represents the height of the reference plane at the i-th position of the X-axis and the j-th position of the Y-axis. ij3 H represents the vertical distance between the data point and the ground plane. ij4 This represents the vertical distance between the data point and the reference plane.

[0065] Calculate the slope between adjacent data points to be tested, and find the corresponding fastener type by referring to the fastener design specification;

[0066]

[0067] Where Yij is the horizontal position of the data point at position ij, and G ij1 G is the slope between the data point and its previous neighbor. ij2 This represents the slope between the data point and its next adjacent point.

[0068] The present invention also provides a rail fastener inspection system, comprising:

[0069] First scanning module: used to scan the track sequentially using lidar to obtain a three-dimensional point cloud map of the track;

[0070] Denoising module: used to denoise the 3D point cloud image using pass-through filtering and voxel filtering to reduce the number of data points in the 3D point cloud image;

[0071] Calculation module: used to take the highest point at each location of the three-dimensional point cloud map as a reference plane to calculate the corresponding ground plane;

[0072] The first analysis module is used to perform feature analysis on the three-dimensional point cloud map to find the point cloud representing the rail, so as to obtain the rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other;

[0073] The second analysis module is used to analyze the vertical and horizontal positions of the rail point cloud blocks and identify interfering point cloud blocks that do not represent the rail body itself.

[0074] The second scanning module is used to scan the track using a track inspection instrument to obtain a chord data map. Based on the chord data map, it identifies the interference point cloud blocks representing fasteners and uses them as measurement point cloud blocks.

[0075] Determination module: used to calculate the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, so as to determine the type of fastener;

[0076] Monitoring module: used to periodically monitor the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, in order to detect the integrity of each fastener.

[0077] The present invention also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the rail fastener detection method described above.

[0078] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the rail fastener detection method as described above.

[0079] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0080] By scanning the track with lidar, a 3D point cloud map of the track area can be obtained. Then, the reference plane and ground plane of the 3D point cloud map are identified vertically. At this stage, the 3D point cloud map also includes ballast, weeds, etc., near the rails. Therefore, feature analysis is performed on the 3D point cloud map. First, data points represented by ballast and weeds are excluded, resulting in rail point cloud blocks. These rail point cloud blocks also contain data points representing fasteners and foreign objects on the rail. Therefore, by analyzing the vertical and horizontal positions of the rail point cloud blocks and referring to the dimensions designed for the rails, fasteners that do not meet the design dimensions can be removed. Foreign objects are considered interfering point cloud blocks. Due to the large number of data points from fasteners and foreign objects, a track inspection instrument is used to scan the track to obtain chord data maps, thus eliminating foreign objects other than fasteners and obtaining measurement point cloud blocks. The positional relationship between the data points in the measurement point cloud blocks and the reference plane and the ground plane, as well as the positional relationship between the data points in the measurement point cloud blocks, are analyzed. Referring to the dimensions of the fasteners during design, the type of fastener is determined. Subsequently, it is only necessary to analyze the missing data points in the measurement point cloud blocks to know the damage status of the fasteners, thereby enabling accurate and rapid inspection of the fasteners.

[0081] This invention uses point cloud methods to determine the location of damage to fasteners. Compared with visual solutions such as image processing, this solution can greatly reduce the interference of the environment on the accuracy of recognition. Attached Figure Description

[0082] Figure 1 This is a flowchart of the rail fastener inspection method in the first embodiment of the present invention;

[0083] Figure 2 This is a schematic diagram of the U-shaped fastener in the first embodiment of the present invention;

[0084] Figure 3 This is a structural block diagram of the rail fastener detection system in the second embodiment of the present invention;

[0085] Figure 4 This is a structural block diagram of the electronic device in the third embodiment of the present invention;

[0086] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. Detailed Implementation

[0087] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0088] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0089] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0090] Please see Figure 1 The image shows a method for detecting rail fasteners according to the first embodiment of the present invention, which includes the following steps:

[0091] S10, uses lidar to scan the track sequentially to obtain a three-dimensional point cloud map of the track;

[0092] S20, taking the highest point at each location of the three-dimensional point cloud map as a reference plane to calculate the corresponding ground plane;

[0093] S30, Perform feature analysis on the three-dimensional point cloud map to find the point cloud representing the rail, so as to obtain the rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other;

[0094] S40, Analyze the vertical and horizontal positions of the rail point cloud blocks to identify interfering point cloud blocks that do not represent the rail body.

[0095] S50, the track is scanned by a track inspection instrument to obtain a chord data map. Based on the chord data map, the interference point cloud blocks representing the fasteners are identified and used as measurement point cloud blocks.

[0096] S60, calculate the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, to determine the type of fastener;

[0097] S70, periodically monitor the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane to detect the integrity of each fastener.

[0098] Understandably, this invention uses laser radar to scan the track to obtain a 3D point cloud map of the track. Then, it identifies the reference plane and the ground plane of the 3D point cloud map in a vertical plane. At this point, the 3D point cloud map also includes ballast, weeds, etc., near the rails. Therefore, feature analysis is performed on the 3D point cloud map to first exclude data points represented by ballast and weeds, resulting in rail point cloud blocks. These rail point cloud blocks also contain data points represented by fasteners and foreign objects on the rails. Therefore, by analyzing the vertical and horizontal positions of the rail point cloud blocks and referring to the dimensions designed for the rails, data points that do not meet the design specifications can be removed. Fasteners and foreign objects of varying sizes are considered interfering point cloud blocks. Due to the large number of data points from fasteners and foreign objects, a track inspection instrument is used to scan the track to obtain chord data maps, thus eliminating foreign objects other than fasteners and obtaining measurement point cloud blocks. The positional relationship between the data points in the measurement point cloud blocks and the reference plane and the ground plane, as well as the positional relationship between the data points in the measurement point cloud blocks, are analyzed. Referring to the dimensions of the fasteners during design, the type of fastener is determined. Subsequently, it is only necessary to analyze the missing data points in the measurement point cloud blocks to know the damage status of the fasteners, thereby enabling accurate and rapid inspection of the fasteners.

[0099] This invention uses point cloud methods to determine the location of damage to fasteners. Compared with visual solutions such as image processing, this solution can greatly reduce the interference of the environment on the accuracy of recognition.

[0100] Specifically, in this embodiment, after step S10, the following is also included:

[0101] The 3D point cloud image is denoised using direct-pass filtering and voxel filtering to reduce the number of data points in the 3D point cloud image.

[0102] Understandably, the pass-through filtering method defines a range on three axes to perform preliminary filtering on data points within the specified range in the 3D point cloud image. For example, the location of weeds usually has a preliminary range. We set the preliminary range representing weeds as the range to be filtered by the pass-through filtering method for coarse filtering. The voxel filtering method downsamples the 3D point cloud image to reduce the number of data points and increase the processing speed of data points.

[0103] Furthermore, the specific steps of step S20 include:

[0104] At the highest point of the three-dimensional point cloud map, three points are randomly selected to construct multiple reference planes. Each data point on the reference plane is extracted as the top surface data point, and all data points in the vertical direction of the top surface data point are taken as undetermined data points.

[0105] Set a first height threshold, and calculate the first vertical height from each of the top surface data points to each of the undetermined data points in its vertical direction;

[0106] Calculate whether the first vertical height is greater than the first height threshold. If so, then take the undetermined data point as a ground data point.

[0107] An optimization formula is introduced to calculate whether the number of ground data points meets the predetermined number based on the current environment of the orbit.

[0108] If the predicted probability is less than the preset probability, then three points are reselected to construct the reference plane, and the number of ground data points is iteratively calculated until the predicted probability is greater than the preset probability.

[0109]

[0110] Where P represents the prediction probability, γ represents the number of predetermined ground data points, θ represents the number of ground data points that are exceeded or missing, q represents the number of all undetermined data points in the vertical direction, and Q represents the number of iterations.

[0111] The plane formed by the highest ground data points obtained in the last calculation is taken as the ground surface plane.

[0112] Understandably, in a 3D point cloud map, the highest point along the Z-axis is generally the upper surface of the rail. Since the rail's height is fixed during design, by arbitrarily selecting three points along the rail's extension direction from the highest point in the 3D point cloud map (i.e., the point on the upper surface of the rail), multiple reference planes can be constructed. Each reference plane is represented by a single plane equation. Points on these reference planes are called top surface data points. Then, the first vertical height from the top surface data point to all undetermined data points below is calculated. If this first vertical height is greater than a first height threshold (the first height threshold is the vertical height of the rail), then the undetermined data point is considered a ground data point. To avoid environmental interference... The number of data points can have an impact. For example, if this location happens to be a weld seam, or if there are foreign objects on the rail, one of the points in the reference plane constructed from the three selected points may be higher, resulting in a larger number of ground data points. Alternatively, if there are defects in the rail, one of the points in the reference plane may be lower, resulting in a smaller number of ground data points. Therefore, it is necessary to redetermine the reference plane and calculate whether the number of ground data points meets the predetermined number, until the predicted probability in the optimization formula is greater than the preset probability. The plane formed by the highest ground data points obtained in the last calculation is then taken as the ground surface plane. This determines the upper and lower planes of the rail.

[0113] Furthermore, the specific steps of step S30 include:

[0114] Using the ground surface as a reference plane and the extension direction of the rails in the track as the X-axis, multiple starting points x in the X-axis are proposed. n And set a sliding window (x n ,0);

[0115] Based on the preset width of the rail as the search command, the sliding window is slid along the Y-axis of the rail width direction to obtain the bottom window pane point of the rail;

[0116] (w 11 w 22 w 33 ......w nm )

[0117] Among them, w nm This indicates starting from the nth starting point x. n The m-th bottom window pane point of the starting rail;

[0118] All data points above the window panes on the bottom surface of the rail are extracted and regarded as rail body data points to obtain a rail point cloud map. Then, two rail point cloud blocks are extracted from the rail point cloud map.

[0119] Understandably, after calculating the reference plane and the ground plane, it is necessary to obtain the actual position of the rail in the 3D point cloud map. By using the ground plane as the reference plane, setting multiple starting points along the extension direction of the rail, using the vertical direction of the rail as the sliding direction, and using the width of the rail as the search command, the bottom window points of the rail are found. Since the dimensions of each section of the rail are not consistent, all window points on the bottom window points of the rail are regarded as window points of the rail body, thus obtaining the rail point cloud map and two rail point cloud blocks.

[0120] Furthermore, the specific steps of step S40 include:

[0121] Calculate the second vertical height of each data point of the rail point cloud block with its left and right adjacent points;

[0122] H ij1 =|z ij -z i(j+1) |

[0123] H ij2 =|z ij -z i(j-1) |;

[0124] Among them, Z ij H represents the height of the data point at position i on the X-axis and position j on the Y-axis. ij1 H is the second vertical height between the data point and its right neighbor. ij2 The second vertical height between the data point and its left adjacent point;

[0125] A second height threshold is set. If the second vertical height is within the range of the second height threshold, then the left adjacent point or the right adjacent point is regarded as a ballast point cloud block to be determined.

[0126] Import historical data from the undetermined ballast point cloud block and perform a reliability assessment on the undetermined ballast point cloud block;

[0127]

[0128] Where t is time, k is the number of measurements, and Z is the time interval. tij Let be the height of the data point at position ij at time t. To measure the average height of data points at position ij within a time range, δ ij Let be the standard deviation of the height of the data point at position ij;

[0129] If the data points in the undetermined ballast point cloud block meet the credibility judgment criteria, then the undetermined ballast point cloud block is regarded as a ballast point cloud block, and the ballast point cloud block is excluded from the rail point cloud block.

[0130]

[0131] Where μ is the error range, 2≤μ≤3;

[0132] In the updated rail point cloud block, the horizontal distance between the edge data points at each horizontal height is calculated. Referring to the rail design standard, edge data points whose horizontal distance is greater than the rail design standard are extracted and regarded as undetermined foreign object point cloud blocks.

[0133] Import historical data of the point cloud block of the undetermined foreign object, and evaluate the credibility of the point cloud block of the undetermined foreign object;

[0134]

[0135] Among them, X tij Let be the vertical horizontal position of the data point at position ij at time t. To measure the average vertical horizontal position of data points at position ij within the time range, Δ ij Let be the standard deviation of the vertical horizontal position of the data point at position ij;

[0136] If the data points in the undetermined foreign object point cloud block meet the credibility judgment criteria, then the undetermined foreign object point cloud block is regarded as a foreign object point cloud block.

[0137]

[0138] The ballast point cloud block and the foreign object point cloud block are considered together as interference point cloud blocks.

[0139] Understandably, the bottom of the rail is very close to the ground, but there is still a certain distance. There is ballast next to the rail, which is very close to the rail and even extends to the web of the rail. In order to eliminate the interference of ballast on the rail point cloud block, the second vertical height of each data point in the rail point cloud block with its left and right adjacent points is calculated. It is also calculated whether the second vertical height with the left and right adjacent points is within the second height threshold (generally the distance from the lower flange to the road surface and the distance from the upper flange to the lower flange). Generally speaking, the data points in the rail point cloud block are continuous with their left and right adjacent points. If there is a sudden change in height, it means that the data points have encountered foreign objects or ballast, or the upper flange has fallen to the lower flange, or the lower flange has fallen to the road surface. If the data points are not within the second height threshold but a sudden change in height occurs, it means that the data points have encountered foreign objects or ballast and should be removed.

[0140] Since it is common for ballast to spill onto the rails, removing these data points would be too much work. Therefore, historical data is introduced to examine the value of the second vertical height at that location over a period of time. If the data does not meet the reliability criteria, it means that the ballast at that location has fallen onto the rails due to environmental reasons, or that the ballast at that location has become ballast at a different height (indicating that the rails at that location are prone to ballast accumulation). In this case, there is no need to remove it, as it will fall off over time. If the data meets the reliability criteria, it means that the ballast at that location is very solid and needs to be removed, thus eliminating ballast point cloud blocks.

[0141] Since the distance between the cross sections at various heights of the rail is determined by the rail design standard, if the horizontal distance of a cross section is greater than the rail design standard, it means that there is a foreign object. According to the above-mentioned ballast reliability assessment steps, historical data is introduced to assess the reliability of the foreign object. If it does not meet the reliability judgment standard, it means that the foreign object is easy to fall off and does not need to be ignored. If it meets the reliability judgment standard, it should be regarded as a foreign object point cloud block and extracted. Then, the ballast point cloud block and the foreign object point cloud block are considered together as interference point cloud blocks.

[0142] Furthermore, the specific steps of step S50 include:

[0143] The track is scanned using a track inspection instrument to obtain a chord data map, and a chord data matrix T is constructed based on the chord data map;

[0144]

[0145] Where N represents the total number of scans by the track inspection instrument. For scanning and sampling data;

[0146] Based on the matrix binary recursive construction method, the chord data matrix is ​​decomposed by singular value decomposition to obtain similar singular values ​​and key singular values, and then the similar signals ε and key signals λ corresponding to the similar singular values ​​and key singular values ​​are obtained respectively.

[0147] The similar signals are decomposed step by step to obtain the similar signals ε at each level. a and the key signal λ a Where a is the number of factorizations;

[0148] Construct a key signal matrix D based on the key signals at all levels. a ;

[0149] In the key signal matrix D a Search for signals with abrupt peak changes and consider these signals as the first undetermined signals.

[0150] Analyze signals with the same peaks and consistent spacing in the first undetermined signal, and regard them as the second undetermined signal;

[0151] Referring to the fastener design specifications, the second undetermined signal whose spacing conforms to the fastener design specifications is regarded as the fastener signal, and the interference point cloud block at the position of the fastener signal is regarded as the measurement point cloud block.

[0152] Understandably, since it is difficult to accurately distinguish between foreign objects and fasteners from point cloud maps, a track inspection instrument is introduced to scan the track to obtain chord data maps, and then construct a chord data matrix H. The chord data matrix H is then subjected to singular value decomposition (SVD) based on the matrix binary recursive construction method. Since the chord signal will show peaks (i.e., singular values) when encountering foreign objects, and also when the rail is uneven, to distinguish whether the peaks are caused by unevenness, foreign objects, or fasteners, the chord data matrix H is subjected to SVD. This yields similar singular values ​​and critical singular values, and thus similar signals and critical signals. Similar signals are generally caused by unevenness and are not intense, so they look similar. Critical signals reflect peaks caused by foreign objects, fasteners, etc. Through progressive decomposition, purer critical signals can be obtained, resulting in a critical signal matrix D. j It can be derived from the key signal matrix D j The signal that indicates a sudden change in peak is searched. The peaks reflected by foreign objects have a characteristic that the peaks of adjacent foreign objects are inconsistent and the spacing is not equal. Therefore, such signals can be eliminated. The peaks caused by welds and fasteners have a characteristic that the peaks are consistent and the spacing is equal. Therefore, according to the spacing specifications between fasteners in the fastener design specifications, we find the signals that represent fasteners and eliminate the signals that represent welds. Then, the interfering point cloud blocks at the same position are regarded as the measurement point cloud blocks.

[0153] Furthermore, the specific steps of step S60 include:

[0154] Calculate the vertical distance between each data point in the measurement point cloud block and the reference plane and the ground surface plane respectively. Referring to the fastener design specification, data points whose vertical distance conforms to the fastener design specification are regarded as data points to be verified.

[0155] H ij3 =z ij

[0156] H ij4 =|z ij -z′ ij |;

[0157] Where, z′ ij H represents the height of the reference plane at the i-th position of the X-axis and the j-th position of the Y-axis.ij3 H represents the vertical distance between the data point and the ground plane. ij4 This represents the vertical distance between the data point and the reference plane.

[0158] Calculate the slope between adjacent data points to be tested, and find the corresponding fastener type by referring to the fastener design specification;

[0159]

[0160] Where Yij is the horizontal position of the data point at position ij, and G ij1 G is the slope between the data point and its previous neighbor. ij2 This represents the slope between the data point and its next adjacent point.

[0161] Understandably, taking a U-shaped fastener as an example, its shape is like... Figure 2 As shown, the fastener spring is U-shaped, and the fastener bolt does not contact the spring. We calculate the vertical distances between each data point in the measurement point cloud representing the U-shaped fastener and the reference plane and the ground plane, the vertical distances between the bottom edge of the U-shape and the ground plane, the vertical distances between the top edge of the U-shape and the reference plane, and the vertical distances between the right edge of the U-shape and the reference plane and the ground plane. Based on the vertical distances between each data point and the reference plane and the ground plane, we can initially consider that the measurement point cloud represents a U-shaped fastener and regard it as a data point to be verified. Then, we focus on calculating the slopes between the right edge of the U-shape and the bottom and top edges of the U-shape, especially at the corners. This can basically confirm that the fastener type is a U-shaped fastener.

[0162] Afterwards, by repeating steps S10 to S60 periodically, the measurement point cloud block can be updated. Based on the missing or added data points of each data point in the measurement point cloud block (foreign objects on the fastener), the integrity of each fastener can be detected, thereby enabling comprehensive monitoring of the fastener.

[0163] In summary, the rail fastener detection method in the above embodiments of the present invention determines the location of damage to the fastener through point cloud analysis. Compared with visual solutions such as image processing, this solution can greatly reduce the interference of the environment on the recognition accuracy.

[0164] Please refer to Figure 3 The image shows a rail fastener inspection system according to a second embodiment of the present invention, comprising:

[0165] First scanning module 11: Used to scan the track sequentially with a lidar to obtain a three-dimensional point cloud map of the track;

[0166] Denoising module 12: used to denoise the three-dimensional point cloud map using a pass-through filtering method and a voxel filtering method to reduce the number of data points in the three-dimensional point cloud map;

[0167] Calculation module 13: used to take the highest point of each location in the three-dimensional point cloud map as a reference plane to calculate the corresponding ground plane;

[0168] The calculation module 13 is specifically used for:

[0169] At the highest point of the three-dimensional point cloud map, three points are randomly selected to construct multiple reference planes. Each data point on the reference plane is extracted as the top surface data point, and all data points in the vertical direction of the top surface data point are taken as undetermined data points.

[0170] Set a first height threshold, and calculate the first vertical height from each of the top surface data points to each of the undetermined data points in its vertical direction;

[0171] Calculate whether the first vertical height is greater than the first height threshold. If so, then take the undetermined data point as a ground data point.

[0172] An optimization formula is introduced to calculate whether the number of ground data points meets the predetermined number based on the current environment of the orbit.

[0173] If the predicted probability is less than the preset probability, then three points are reselected to construct the reference plane, and the number of ground data points is iteratively calculated until the predicted probability is greater than the preset probability.

[0174]

[0175] Where P represents the prediction probability, γ represents the number of predetermined ground data points, θ represents the number of ground data points that are exceeded or missing, q represents the number of all undetermined data points in the vertical direction, and Q represents the number of iterations.

[0176] The plane formed by the highest ground data point obtained in the last calculation is taken as the ground surface plane;

[0177] First analysis module 14: used to perform feature analysis on the three-dimensional point cloud map, find the point cloud representing the rail, and obtain the rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other;

[0178] The first analysis module 14 is specifically used for:

[0179] Using the ground surface as a reference plane and the extension direction of the rails in the track as the X-axis, multiple starting points x in the X-axis are proposed. n And set a sliding window (x n ,0);

[0180] Based on the preset width of the rail as the search command, the sliding window is slid along the Y-axis of the rail width direction to obtain multiple rail bottom window points;

[0181] (w 11 w 22 w 33 ......w nm )

[0182] Among them, w nm This indicates starting from the nth starting point x. n The m-th bottom window pane point of the starting rail;

[0183] All data points above the window panes on the bottom surface of the rail are extracted and regarded as rail body data points to obtain a rail point cloud map. Then, two rail point cloud blocks are extracted from the rail point cloud map.

[0184] The second analysis module 15 is used to analyze the vertical and horizontal positions of the rail point cloud blocks and identify interfering point cloud blocks that do not represent the rail body.

[0185] The second analysis module 15 is specifically used for:

[0186] Calculate the second vertical height of each data point of the rail point cloud block with its left and right adjacent points;

[0187] H ij1 =|z ij -z i(j+1) |

[0188] H ij2 =|z ij -z i(j-1) |;

[0189] Among them, Z ij H represents the height of the data point at position i on the X-axis and position j on the Y-axis. ij1 H is the second vertical height between the data point and its right neighbor. ij2 The second vertical height between the data point and its left adjacent point;

[0190] A second height threshold is set. If the second vertical height is within the range of the second height threshold, then the left adjacent point or the right adjacent point is regarded as a ballast point cloud block to be determined.

[0191] Import historical data from the undetermined ballast point cloud block and perform a reliability assessment on the undetermined ballast point cloud block;

[0192]

[0193] Where t is time, k is the number of measurements, and Z is the time interval.tij Let be the height of the data point at position ij at time t. To measure the average height of data points at position ij within a time range, δ ij Let be the standard deviation of the height of the data point at position ij;

[0194] If the data points in the undetermined ballast point cloud block meet the credibility judgment criteria, then the undetermined ballast point cloud block is regarded as a ballast point cloud block, and the ballast point cloud block is excluded from the rail point cloud block.

[0195]

[0196] Where μ is the error range, 2≤μ≤3;

[0197] In the updated rail point cloud block, the horizontal distance between the edge data points at each horizontal height is calculated. Referring to the rail design standard, edge data points whose horizontal distance is greater than the rail design standard are extracted and regarded as undetermined foreign object point cloud blocks.

[0198] Import historical data of the point cloud block of the undetermined foreign object, and evaluate the credibility of the point cloud block of the undetermined foreign object;

[0199]

[0200] Among them, X tij Let be the vertical horizontal position of the data point at position ij at time t. To measure the average vertical horizontal position of data points at position ij within the time range, Δ ij Let be the standard deviation of the vertical horizontal position of the data point at position ij;

[0201] If the data points in the undetermined foreign object point cloud block meet the credibility judgment criteria, then the undetermined foreign object point cloud block is regarded as a foreign object point cloud block.

[0202]

[0203] The ballast point cloud block and the foreign object point cloud block are collectively regarded as interference point cloud blocks;

[0204] The second scanning module 16 is used to scan the track with a track inspection instrument to obtain a chord data map, and based on the chord data map, to find the interference point cloud blocks representing the fasteners and use them as measurement point cloud blocks.

[0205] The second scanning module 16 is specifically used for:

[0206] The track is scanned using a track inspection instrument to obtain a chord data map, and a chord data matrix T is constructed based on the chord data map;

[0207]

[0208] Where N represents the total number of scans by the track inspection instrument. For scanning and sampling data;

[0209] Based on the matrix binary recursive construction method, the chord data matrix is ​​decomposed by singular value decomposition to obtain similar singular values ​​and key singular values, and then the similar signals ε and key signals λ corresponding to the similar singular values ​​and key singular values ​​are obtained respectively.

[0210] The similar signals are decomposed step by step to obtain the similar signals ε at each level. a and the key signal λ a Where a is the number of factorizations;

[0211] Construct a key signal matrix D based on the key signals at all levels. a ;

[0212] In the key signal matrix D a Search for signals with abrupt peak changes and consider these signals as the first undetermined signals.

[0213] Analyze signals with the same peaks and consistent spacing in the first undetermined signal, and regard them as the second undetermined signal;

[0214] Referring to the fastener design specifications, the second undetermined signal whose spacing conforms to the fastener design specifications is regarded as a fastener signal, and the interference point cloud block at the position of the fastener signal is regarded as the measurement point cloud block;

[0215] Determining module 17: Calculates the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, in order to determine the type of fastener;

[0216] The determining module 17 is specifically used for:

[0217] Calculate the vertical distance between each data point in the measurement point cloud block and the reference plane and the ground surface plane respectively. Referring to the fastener design specification, data points whose vertical distance conforms to the fastener design specification are regarded as data points to be verified.

[0218] H ij3 =z ij

[0219] H ij4 =|z ij -z′ ij |;

[0220] Where, z′ij H represents the height of the reference plane at the i-th position of the X-axis and the j-th position of the Y-axis. ij3 H represents the vertical distance between the data point and the ground plane. ij4 This represents the vertical distance between the data point and the reference plane.

[0221] Calculate the slope between adjacent data points to be tested, and find the corresponding fastener type by referring to the fastener design specification;

[0222]

[0223] Where Yij is the horizontal position of the data point at position ij, and G ij1 G is the slope between the data point and its previous neighbor. ij2 The slope between the data point and its next adjacent point;

[0224] Monitoring module 18: used to periodically monitor the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, in order to detect the integrity of each fastener.

[0225] The present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 4 The diagram shows an electronic device according to a third embodiment of the present invention, including a memory 10, a processor 20, and a computer program 30 stored in the memory 10 and executable on the processor 20. When the processor 20 executes the computer program 30, it implements the above-described rail fastener detection method.

[0226] The memory 10 includes at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 10 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 10 can be an external storage device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Furthermore, the memory 10 can include both internal and external storage units of the electronic device. The memory 10 can be used not only to store application software and various types of data installed in the electronic device, but also to temporarily store data that has been output or will be output.

[0227] In some embodiments, the processor 20 may be an electronic control unit (ECU, also known as a vehicle computer), a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip, used to run program code stored in the memory 10 or process data, such as executing access restriction programs.

[0228] It should be pointed out that, Figure 4 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0229] This invention also proposes a readable storage medium storing a computer program that, when executed by a processor, implements the rail fastener detection method described above.

[0230] Those skilled in the art will understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequential list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0231] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0232] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0233] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0234] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for inspecting rail fasteners, characterized in that, The steps include: The track is scanned sequentially using a lidar to obtain a three-dimensional point cloud map of the track. The highest point at each location in the three-dimensional point cloud map is taken as the reference plane to calculate the corresponding ground plane; Feature analysis is performed on the three-dimensional point cloud map to identify the point cloud representing the rail, thus obtaining a rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other; The vertical and horizontal positions of the rail point cloud blocks are analyzed to identify interfering point cloud blocks that do not represent the rail body itself. The track is scanned using a track inspection instrument to obtain a chord data map. Based on the chord data map, the interference point cloud blocks representing the fasteners are identified and used as the measurement point cloud blocks. The geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, are calculated to determine the type of fastener; The geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, are monitored periodically to detect the integrity of each fastener; The step of taking the highest point at each location in the three-dimensional point cloud map as a reference plane to calculate the corresponding ground plane specifically includes: At the highest point of the three-dimensional point cloud map, three points are randomly selected to construct multiple reference planes. Each data point on the reference plane is extracted as the top surface data point, and all data points in the vertical direction of the top surface data point are taken as undetermined data points. Set a first height threshold, and calculate the first vertical height from each of the top surface data points to each of the undetermined data points in its vertical direction; Calculate whether the first vertical height is greater than the first height threshold. If so, then take the undetermined data point as a ground data point. An optimization formula is introduced to calculate whether the number of ground data points meets the predetermined number based on the current environment of the orbit. If the predicted probability is less than the preset probability, then three points are reselected to construct the reference plane, and the number of ground data points is iteratively calculated until the predicted probability is greater than the preset probability. ； Where P represents the prediction probability, γ represents the number of predetermined ground data points, θ represents the number of ground data points that are exceeded or missing, q represents the number of all undetermined data points in the vertical direction, and Q represents the number of iterations. The plane formed by the highest ground data point obtained in the last calculation is taken as the ground surface plane; The analysis of the vertical and horizontal positions of the rail point cloud blocks to identify interfering point cloud blocks that do not represent the rail body specifically includes: Calculate the second vertical height of each data point of the rail point cloud block with its left and right adjacent points; Among them, Z ij H represents the height of the data point at position i on the X-axis and position j on the Y-axis. ij1 H is the second vertical height between the data point and its right neighbor. ij2 The second vertical height between the data point and its left adjacent point; A second height threshold is set. If the second vertical height is within the range of the second height threshold, then the left adjacent point or the right adjacent point is regarded as a ballast point cloud block to be determined. Import historical data from the undetermined ballast point cloud block and perform a reliability assessment on the undetermined ballast point cloud block; Where t is time, k is the number of measurements, and Z is the time interval. tij Let be the height of the data point at position ij at time t. To measure the average height of data points at position ij within a time range, δ ij Let be the standard deviation of the height of the data point at position ij; If the data points in the undetermined ballast point cloud block meet the credibility judgment criteria, then the undetermined ballast point cloud block is regarded as a ballast point cloud block, and the ballast point cloud block is excluded from the rail point cloud block. ； Where μ is the error range, 2≤μ≤3; In the updated rail point cloud block, the horizontal distance between the edge data points at each horizontal height is calculated. Referring to the rail design standard, edge data points whose horizontal distance is greater than the rail design standard are extracted and regarded as undetermined foreign object point cloud blocks. Import historical data of the point cloud block of the undetermined foreign object, and evaluate the credibility of the point cloud block of the undetermined foreign object; Where t is time, k is the number of measurements, and X tij Let be the vertical horizontal position of the data point at position ij at time t. To measure the average vertical horizontal position of the data points at position ij within the time range, Let be the standard deviation of the vertical horizontal position of the data point at position ij; If the data points in the undetermined foreign object point cloud block meet the credibility judgment criteria, then the undetermined foreign object point cloud block is regarded as a foreign object point cloud block. ； Where μ is the error range, 2≤μ≤3; The ballast point cloud block and the foreign object point cloud block are considered together as interference point cloud blocks.

2. The rail fastener inspection method according to claim 1, characterized in that, The process of sequentially scanning the track with a lidar to obtain a three-dimensional point cloud map of the track location, followed by: The 3D point cloud image is denoised using direct-pass filtering and voxel filtering to reduce the number of data points in the 3D point cloud image.

3. The rail fastener inspection method according to claim 1, characterized in that, The feature analysis of the three-dimensional point cloud map is performed to identify the point cloud representing the rail location, thus obtaining a rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other, specifically including: Using the ground surface as a reference plane and the extension direction of the rails in the track as the X-axis, multiple starting points are identified along the X-axis. x n And set a sliding window ( x n ,0); Based on the preset width of the rail as the search command, the sliding window is slid along the Y-axis of the rail width direction to obtain multiple rail bottom window points; ； Among them, w nm Indicates starting from the nth point x n The m-th bottom window pane point of the starting rail; All data points above the window panes on the bottom surface of the rail are extracted and regarded as rail body data points to obtain a rail point cloud map. Then, two rail point cloud blocks are extracted from the rail point cloud map.

4. The rail fastener inspection method according to claim 1, characterized in that, The process involves scanning the track using a track inspection instrument to obtain a chord data map. Based on this chord data map, interference point cloud blocks representing fasteners are identified and used as measurement point cloud blocks. Specifically, this includes: The track is scanned using a track inspection instrument to obtain a chord data map, and a chord data matrix T is constructed based on the chord data map; ； Where N represents the total number of scans by the track inspection instrument. For scanning and sampling data; Based on the matrix binary recursive construction method, the chord data matrix is ​​decomposed by singular value decomposition to obtain similar singular values ​​and key singular values, and then the similar signals ε and key signals λ corresponding to the similar singular values ​​and key singular values ​​are obtained respectively. The similar signals are decomposed step by step to obtain the similar signals ε at each level. a and the key signal λ a Where a is the number of factorizations; Construct a key signal matrix D based on the key signals at all levels. a ; In the key signal matrix D a Search for signals with abrupt peak changes and consider these signals as the first undetermined signals. Analyze signals with the same peaks and consistent spacing in the first undetermined signal, and regard them as the second undetermined signal; Referring to the fastener design specifications, the second undetermined signal whose spacing conforms to the fastener design specifications is regarded as the fastener signal, and the interference point cloud block at the position of the fastener signal is regarded as the measurement point cloud block.

5. The rail fastener inspection method according to claim 4, characterized in that, The calculation of the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, to determine the type of fastener, specifically includes: Calculate the vertical distance between each data point in the measurement point cloud block and the reference plane and the ground surface plane respectively. Referring to the fastener design specification, data points whose vertical distance conforms to the fastener design specification are regarded as data points to be verified. in, Z represents the height of the reference plane at the i-th position of the X-axis and the j-th position of the Y-axis. ij H represents the height of the data point at position i on the X-axis and position j on the Y-axis. ij3 H represents the vertical distance between the data point and the ground plane. ij4 This represents the vertical distance between the data point and the reference plane. Calculate the slope between adjacent data points to be tested, and find the corresponding fastener type by referring to the fastener design specification; Among them, Z ij Let X be the height of the data point at position i on the X-axis and position j on the Y-axis. ij Y represents the vertical horizontal position of the data point at position ij. ij Let G be the horizontal position of the data point at position ij. ij1 G is the slope between the data point and its previous neighbor. ij2 This represents the slope between the data point and its next adjacent point.

6. A rail fastener inspection system, used to implement the rail fastener inspection method according to any one of claims 1 to 5, characterized in that, include: First scanning module: used to scan the track sequentially using lidar to obtain a three-dimensional point cloud map of the track; Calculation module: used to take the highest point at each location of the three-dimensional point cloud map as a reference plane to calculate the corresponding ground plane; The calculation module is specifically used for: At the highest point of the three-dimensional point cloud map, three points are randomly selected to construct multiple reference planes. Each data point on the reference plane is extracted as the top surface data point, and all data points in the vertical direction of the top surface data point are taken as undetermined data points. Set a first height threshold, and calculate the first vertical height from each of the top surface data points to each of the undetermined data points in its vertical direction; Calculate whether the first vertical height is greater than the first height threshold. If so, then take the undetermined data point as a ground data point. An optimization formula is introduced to calculate whether the number of ground data points meets the predetermined number based on the current environment of the orbit. If the predicted probability is less than the preset probability, then three points are reselected to construct the reference plane, and the number of ground data points is iteratively calculated until the predicted probability is greater than the preset probability. ； Where P represents the prediction probability, γ represents the number of predetermined ground data points, θ represents the number of ground data points that are exceeded or missing, q represents the number of all undetermined data points in the vertical direction, and Q represents the number of iterations. The plane formed by the highest ground data point obtained in the last calculation is taken as the ground surface plane; The first analysis module is used to perform feature analysis on the three-dimensional point cloud map to find the point cloud representing the rail, so as to obtain the rail point cloud map; the rail point cloud map includes two rail point cloud blocks arranged opposite each other; The second analysis module is used to analyze the vertical and horizontal positions of the rail point cloud blocks and identify interfering point cloud blocks that do not represent the rail body itself. The second analysis module is specifically used for: Calculate the second vertical height of each data point of the rail point cloud block with its left and right adjacent points; Among them, Z ij H represents the height of the data point at position i on the X-axis and position j on the Y-axis. ij1 H is the second vertical height between the data point and its right neighbor. ij2 The second vertical height between the data point and its left adjacent point; A second height threshold is set. If the second vertical height is within the range of the second height threshold, then the left adjacent point or the right adjacent point is regarded as a ballast point cloud block to be determined. Import historical data from the undetermined ballast point cloud block and perform a reliability assessment on the undetermined ballast point cloud block; ； Where t is time, k is the number of measurements, and Z is the time interval. tij Let be the height of the data point at position ij at time t. To measure the average height of data points at position ij within a time range, δ ij Let be the standard deviation of the height of the data point at position ij; If the data points in the undetermined ballast point cloud block meet the credibility judgment criteria, then the undetermined ballast point cloud block is regarded as a ballast point cloud block, and the ballast point cloud block is excluded from the rail point cloud block. ； Where μ is the error range, 2≤μ≤3; In the updated rail point cloud block, the horizontal distance between the edge data points at each horizontal height is calculated. Referring to the rail design standard, edge data points whose horizontal distance is greater than the rail design standard are extracted and regarded as undetermined foreign object point cloud blocks. Import historical data of the point cloud block of the undetermined foreign object, and evaluate the credibility of the point cloud block of the undetermined foreign object; Where t is time, k is the number of measurements, and X tij Let be the vertical horizontal position of the data point at position ij at time t. To measure the average vertical horizontal position of the data points at position ij within the time range, Let be the standard deviation of the vertical horizontal position of the data point at position ij; If the data points in the undetermined foreign object point cloud block meet the credibility judgment criteria, then the undetermined foreign object point cloud block is regarded as a foreign object point cloud block. ； Where μ is the error range, 2≤μ≤3; The ballast point cloud block and the foreign object point cloud block are collectively regarded as interference point cloud blocks; The second scanning module is used to scan the track using a track inspection instrument to obtain a chord data map. Based on the chord data map, it identifies the interference point cloud blocks representing fasteners and uses them as measurement point cloud blocks. Determination module: Calculates the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, to determine the type of fastener; Monitoring module: used to periodically monitor the geometric positional relationship between each data point in the measurement point cloud block and the reference plane and the ground surface plane, as well as the geometric positional relationship between adjacent data points, in order to detect the integrity of each fastener.

7. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the rail fastener detection method as described in any one of claims 1-5.

8. An electronic 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 rail fastener detection method as described in any one of claims 1-5.

Images