Rail wear detection method, device and system and storage medium

Abstract

The embodiment of the invention discloses a track wear detection method, device and system and a storage medium. The method comprises the following steps: receiving a scanning point set of a to-be-detected track contour, wherein the to-be-detected track contour comprises a wear region, a stable region, an undetected region and an auxiliary region; preprocessing the scanning point set to generate a modeling point set for generating a first model to generate model points; original points participating in matching are determined in the scanning point set according to a preset matching range, and matching weights are arranged based on the distance sequence between the original points and the model points; determining a target point closest to each original point in the model points, and iteratively calculating a rotation matrix and a translation matrix between the original points and the target points based on the matching weight; and calculating rotation translation values of all the scanning points by using the rotation matrix and the translation matrix to obtain transformation points, and confirming wear information based on the transformation points of the wear region and the model points. The track wear detection method provided by the embodiment of the invention achieves the effect of high efficiency of real-time track wear detection.

Description

Technical Field

[0001] The present invention relates to the field of track detection, and in particular to a track wear detection method, device, system and storage medium. Background Art

[0002] In the field of rail defect detection, profile wear detection is a key task. The profile wear detection results provide important measurement data and evaluation criteria for rail wear assessment, maintenance and replacement. The matching algorithm of the existing rail detection solution has many noise points and a long matching time. Summary of the invention

[0003] The embodiments of the present invention provide a rail wear detection method, device, system and storage medium, which achieve the effect of real-time detection of rail wear with high efficiency.

[0004] According to one aspect of the present invention, a track wear detection method is provided, comprising: receiving a scanning point set of a track profile to be measured, the track profile to be measured comprising at least a worn area, a stable area, an uninspected area and an auxiliary area; preprocessing the scanning point set to generate a modeling point set, generating a first model based on the modeling point set, the first model being used to generate model points; determining original points involved in matching in the scanning point set according to a preset matching range, and setting matching weights based on the distance sorting between the original points and the model points; confirming the target point closest to each original point in the model points, and iteratively calculating the rotation matrix and translation matrix between the original point and the target point based on the matching weight; using the rotation matrix and translation matrix finally obtained after the iteration to calculate the rotation and translation values ​​of all scanning points to obtain transformation points, and confirming the wear information based on the transformation points and model points of the wear area.

[0005] Optionally, preprocessing the scanning point set to generate a modeling point set includes: removing noise points of a frame of scanning point set, the noise points at least including non-track points near the scanning breakpoints at the end of the stable area; filling the contour points of the uninspected area according to the standard track, and forcibly mapping the scanning breakpoints to model inflection points; compensating the contour collection points of the auxiliary area according to the standard track; and compensating the collection points of the worn area according to the top length of the standard track.

[0006] Optionally, determining the original points participating in the matching in the scanning point set according to a preset matching range includes: selecting a first number of scanning points in a first length range of the stable area adjacent to the uninspected area as the first original points participating in the matching, and selecting a second number of scanning points in a second length range of the auxiliary area adjacent to the uninspected area as the second original points participating in the matching.

[0007] Optionally, setting the matching weight based on the distance sorting between the original point and the model point includes: setting the corresponding weight of the point with a smaller distance between the original point and the model point and a first proportion threshold as the first weight, and setting the corresponding weight of the point with a larger distance between the original point and the model point and a second proportion threshold as the second weight.

[0008] Optionally, the iterative calculation of the rotation matrix and the translation matrix between the original points and the target points based on the matching weights includes: calculating the weighted centers of the original point set and the target point set respectively based on the matching weights, the original point set being the set of all original points, and the target point set being the set of all target points; calculating the first point set after the original point set is decentralized, and the second point set after the target point set is decentralized based on the weighted centers; calculating the weighted covariance matrix of the first point set and the second point set; SVD decomposing the covariance matrix to calculate the rotation matrix; using the weighted centers of the first point set and the second point set to calculate the translation matrix T; using the rotation matrix R and the translation matrix T to update the original point set, and repeating the steps after determining the original points participating in the matching in the scanning point set according to the preset matching range, until the maximum number of iterations is reached or the average distance threshold is less than the set value.

[0009] Optionally, confirming the wear information based on the transformation points and model points of the wear area includes: calculating the wear depth based on the transformation points and model points of the wear area, and confirming the wear area based on the area of ​​the intersection of the transformation points and model points of the wear area.

[0010] Optionally, the calculation of the wear depth based on the transformation points and model points of the wear area and the confirmation of the wear area based on the area of ​​the intersection of the transformation points and the model points of the wear area include: intercepting a point set 1 of the transformation points in the wear area, intercepting a point set 2 of the model points in the wear area, calculating the maximum distance point in the vertical direction between the two point sets, which is the maximum wear depth, and calculating the area of ​​the intersection between the two point sets, which is the area of ​​the wear area.

[0011] According to another aspect of the present invention, there is provided a track wear detection device, comprising: a data receiving module, for receiving a scanning point set of a track profile to be measured, wherein the track profile to be measured includes at least a worn area, a stable area, an uninspected area and an auxiliary area; a model generating module, for preprocessing the scanning point set to generate a modeling point set, generating a first model based on the modeling point set, and the first model is used to generate model points; a weight confirmation module, for determining the original points involved in the matching in the scanning point set according to a preset matching range, and setting the matching weights based on the distance sorting between the original points and the model points; a parameter iteration module, for confirming the target point closest to each original point in the model points, and iteratively calculating the rotation matrix and translation matrix between the original point and the target point based on the matching weight; a wear calculation module, for calculating the rotation and translation values ​​of all scanning points using the rotation matrix and translation matrix finally obtained after the iteration to obtain transformation points, and confirming the wear information based on the transformation points and model points of the wear area.

[0012] According to another aspect of the present invention, a track wear detection system is provided, comprising: an image acquisition device for generating a scanning point set by scanning a track, and one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned track wear detection method.

[0013] According to another aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, comprising: when the program is executed by a processor, the track wear detection method as described above is implemented.

[0014] The track wear detection method of the present application realizes on-site modeling by generating a first model based on a scanning point set generated on-site, realizes on-site calibration by determining the original points involved in the matching in the scanning point set based on a preset matching range, and realizes fast iterative calculation by iteratively calculating the rotation matrix and translation matrix between the original point and the target point through matching weights. The track wear detection method of the present application allows on-site staff to perform on-site rapid modeling and calibration before the operation, or to perform on-site track model reconstruction and calibration at regular fixed points without the need for development experience, which can enhance anti-interference and facilitate maintenance. In addition, the embodiment of the present invention carries out targeted design at the business level, introduces prior knowledge of on-site operations through a weight confirmation module to set matching weights, effectively reduces the algorithm matching calculation complexity and the number of iterations, and reduces the interference of noise points, thereby improving the algorithm performance and greatly improving the algorithm operation speed, and can support faster scanning speeds, and can further support the scanning equipment to move along the track at high speed to scan and calculate the results at the same time, achieving the effect of high efficiency in real-time detection of track wear. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 This is a flow chart of a rail wear detection method provided in Example 1 of the present invention.

[0017] Figure 2 It is a schematic diagram of the installation of an image acquisition device for acquiring a set of scanning points of a track profile to be measured in one embodiment of the present invention.

[0018] Figure 3 It is a schematic diagram of the distribution of the worn area, stable area, untested area and auxiliary area of ​​the track profile to be tested according to an embodiment of the present invention.

[0019] Figure 4 It is a schematic diagram of a selection range of original points involved in matching according to an embodiment of the present invention.

[0020] Figure 5 It is a schematic diagram of wear information obtained according to an embodiment of the present invention.

[0021] Figure 6 It is a functional module diagram of a track wear detection device provided in Embodiment 2 of the present invention.

[0022] Figure 7 It is a schematic diagram of the framework structure of a track wear detection system provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 only 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 should fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] [Example 1]

[0026] Figure 1 is a flow chart of a rail wear detection method provided in the first embodiment of the present invention. This embodiment can be applied to a rail wear detection device for execution. The device can be implemented by software and / or hardware and can generally be integrated into a rail wear detection system. Figure 1 As shown, the method includes the following operations:

[0027] S110, receiving a scanning point set of a track profile to be measured, wherein the track profile to be measured at least includes a worn area, a stable area, an uninspected area, and an auxiliary area.

[0028] In one embodiment, reference Figure 2 The scanning point set of the track profile to be measured is obtained by an image acquisition device 200, which includes a high-frequency laser line scanning camera 210 arranged at a 45° downward angle toward both sides of the track 220, and the height of the lens of the camera 210 from the track surface is h=25-35cm, for example, 30cm. The high-frequency laser is a laser in the infrared band of 606nm or 808nm, and the sampling frequency of the high-frequency laser is greater than or equal to 20KhZ.

[0029] In one embodiment, reference Figure 3 , the track profile to be measured includes at least a wear area A, a stable area B, an untested area C, an auxiliary area D and an interference area E. In the present embodiment, the wear area A includes the top surface of the track profile to be measured, which is easily crushed or rubbed by the wheels. In the present embodiment, wear usually occurs in this area. Therefore, the wear area A cannot participate in the matching, because if wear occurs, this area will not be consistent with the model, which will result in large iteration errors and time-consuming iterations. The stable area B is the top side that is close to the wheel hub but is not easily crushed or rubbed by the wheel hub. This area is not prone to wear and debris is not allowed to adhere. It is a relatively stable area. The present embodiment is mainly used for matching the model with the scanning point. The untested area C is Figure 2 The area that the middle camera 210 cannot detect due to being blocked by the side corner 211 does not participate in matching or wear assessment and is only used for display. The auxiliary area D is the part of the track profile that is used for support and will not wear. This D area is similar to the B area and is relatively stable. This embodiment is used for auxiliary matching. The interference area E is the part of the bottom of the track profile that is blocked or buried by debris. This part does not participate in matching either.

[0030] S120: Preprocess the scanning point set to generate a modeling point set, and generate a first model based on the modeling point set, where the first model is used to generate model points.

[0031] In one embodiment, preprocessing the scanning point set to generate a modeling point set includes: removing noise points of a frame of scanning point set, wherein the noise points at least include non-track points near the scanning breakpoints at the end of the stable area; filling the contour points of the uninspected area according to the standard track, and forcibly mapping the scanning breakpoints to model inflection points; compensating the contour collection points of the auxiliary area according to the standard track; and compensating the collection points of the worn area according to the top length of the standard track.

[0032] Specifically, after the image acquisition device 200 enters the scene, it first acquires a frame of contour point data, displays the acquired contour point data on a preset display interface, and manually removes the Figure 3 The points (noise) and non-rail points (noise) in the C area in the scan point are detected, and the end of the B area is detected at the same time to confirm that there are no noise points near the breakpoint 221. If there are noise points, the noise points are deleted to eliminate the problem that the camera scan is easily distorted in places with steep gradients. The C area points corresponding to the national standard track are filled in the blank C area of ​​the scanning point. This operation forces the breakpoint 221 of the scanning point to be mapped as the inflection point of the new model (first model). Referring to the tailing length of the E area of ​​the national standard track, the acquisition points are compensated or cut to eliminate the problem of occlusion or coverage on the site or the problem of excessive laser scanning on the site. The compensated acquisition points are used for modeling calibration. Referring to the top length of the A area under the half-track size of the national standard track, the acquisition points are compensated. In this embodiment, the above operations can be performed by pressing buttons and clicking on the display interface preset by the front-end software. Therefore, the staff only needs simple operations to complete the contour modeling (generating the first model) adapted to the site. Even if the camera acquisition state changes under different outdoor temperatures and humidity conditions, the best matching new model (first model) can still be achieved through on-site rapid modeling, thereby greatly improving the anti-interference ability of the system.

[0033] In this embodiment, as long as the points in area B and area D are stable and accurate, the first model can run stably. Although the country has a unified railway track size standard, the standard track point model can be directly used as a benchmark, but in order to improve system stability, maintenance convenience, and algorithm operation speed and accuracy, this embodiment designs a fast modeling module, which only uses some points in areas B and D in the national standard model, combined with the contour points collected on site, to quickly generate a first model adapted to the site.

[0034] S130, determining original points involved in matching in the scan point set according to a preset matching range, and setting matching weights based on the distance ranking between the original points and the model points.

[0035] In one embodiment, the determining of the original points participating in the matching in the scanning point set according to the preset matching range includes: selecting a first number of scanning points in a first length range of the stable area B adjacent to the unchecked area as the first original points participating in the matching, and selecting a second number of scanning points in a second length range of the auxiliary area D adjacent to the unchecked area C as the second original points participating in the matching. Confirm the index value of the inflection point and enter it through the input window on the preset display interface. See also Figure 4 , B1 and D1 are the selection ranges of the original points participating in the matching in the B area and the D area. In one embodiment, the range setting principle is that there should not be too many points participating in the iteration in the B area to prevent the points in the wear area A from entering and causing interference, and there should not be too many points in the D area to prevent the matching algorithm from over-optimizing the D area. In one embodiment, the number of points participating in the matching in the B1 and D1 areas can be kept basically the same. The on-site personnel can complete the calibration by clicking the mouse on the preset display interface to select the above matching range. Specifically, in one embodiment, the B1 area can be 1 / 3-2 / 3 of the length of the B area, for example, it can be 1 / 2; when the number of points collected in the B1 area is less than a preset threshold, for example, 30, no downsampling is performed and all parameters are iterated; when the number of collected points in the B1 area is greater than the preset threshold, for example, 30, 30 points near the breakpoint (inflection point) 221 are completely retained (to prevent downsampling from destroying the accurate position of the inflection point), and the points at other remaining positions of B1 other than the 30 points near the breakpoint 221 are uniformly downsampled. The default parameter can be 6 times downsampling. During the deployment phase, the parameters can be adjusted through trial operation. If it is found that the number of participating points is too large, for example, greater than 45, the number of points can be reduced by increasing the downsampling multiple (for example, to 7-10 times). In actual tests, a total of about 45 iterative matching points in the B1 area have the best effect in terms of performance and speed.

[0036] Further, in the embodiment of the present invention, the length of the D1 region can be 1.5 times the length of the B1 region (because B1 is close to the camera and has a higher point density, in order to ensure that the number of candidate points in the two regions is basically the same, the point selection range of the D1 region is expanded). When the number of acquisitions in the D1 region is less than 30, no downsampling is performed, and all parameter iterations are performed; when the number of acquisition points in the D1 region is greater than the preset threshold, such as 30, 30 points near the breakpoint (inflection point) 221 are completely retained (to prevent downsampling from destroying the accurate position of the inflection point), and the points of the remaining positions of D1 other than the 30 points near the breakpoint 221 are uniformly downsampled. The default parameter can be 6 times downsampling. The parameters can be adjusted and set through trial operation during the deployment phase. If it is found that there are too many participating points, such as more than 45, the number of points can be reduced by increasing the downsampling multiple (for example, increasing it to 7-10 times). In actual tests, the performance and speed of iterative matching points in the D1 region are the best.

[0037] In one embodiment, setting the matching weight based on the distance ranking between the original point and the model point includes: setting the corresponding weight of the point with a smaller distance between the original point and the model point and ranked at the front by a first proportion threshold as the first weight, and setting the corresponding weight of the point with a larger distance between the original point and the model point and ranked at the back by a second proportion threshold as the second weight. In one embodiment, the first proportion threshold may be 70-90%, for example, the second proportion threshold may be 10-30%, the first weight may be 1, and the second weight may be 0. For example, the corresponding weight of the points with a smaller distance between the original point and the model point and ranked at the front by 80% may be set to 1, and the corresponding weight of the points with a larger distance between the original point and the model point and ranked at the back by 20% may be set to 0.

[0038] Specifically, in one embodiment, all scanning points are defined as: The original points involved in the matching in the scan points are defined as: The nearest model point is defined as: The distance between the original point and the model point is calculated as:

[0039] In one embodiment, the distance d of all original points participating in the matching (assuming N points) is calculated and sorted from large to small, and the index value is The distance value is marked as d thresh , the weight function is:

[0040]

[0041] S140, confirming the target point closest to each original point in the model points, and iteratively calculating the rotation matrix and translation matrix between the original point and the target point based on the matching weight.

[0042] In one embodiment, the points closest to the original points are found in the model points and combined into a new point set, which is collectively referred to as the target point below. The calculation formula for the nearest point (Euclidean distance) is:

[0043]

[0044] In one embodiment, the iterative calculation of the rotation matrix and the translation matrix between the original point and the target point based on the matching weights includes:

[0045] S141, calculating the weighted centers of the original point set and the target point set based on the matching weights, respectively, the original point set being the set of all original points, and the target point set being the set of all target points;

[0046] Specifically, the weighted centers of the original point set and the target point set are calculated respectively. In this embodiment, the weighted center of the original point set is calculated as follows: The weighted center of the target point set is calculated as follows:

[0047] S142: Calculate a first point set after the original point set is decentralized based on the weighted center, and a second point set after the target point set is decentralized.

[0048] Among them, the calculation method of the first point set after decentralization is: The calculation method of the second point set after decentralization is:

[0049] S143, calculating the weighted covariance matrix of the first point set and the second point set;

[0050] Specifically, the first point set is defined as: The second point set is defined as: The weight set is defined as: w n×N =diag(w0,w1,w2,w3...w n ). The rotation matrix is ​​calculated as:

[0051] S144, SVD decomposes the covariance matrix and calculates the rotation matrix. The covariance matrix decomposition method is: S = UDV T ; The calculation method of the rotation matrix R is:

[0052] S145, calculating a translation matrix T using the weighted centers of the first point set and the second point set;

[0053] In one embodiment, the translation matrix T is calculated as follows: In this embodiment, the weighted centers of the first point set and the second point set are used for calculation, and calculation cannot be performed through inflection points or breakpoints.

[0054] S146. Use the rotation matrix R and the translation matrix T to update the original point set, and repeat the steps after determining the original points involved in the matching in the scanning point set according to the preset matching range, until the maximum number of iterations is reached or the average distance threshold is less than the set value.

[0055] Specifically, the original point set is updated with the rotation matrix R and the translation matrix T Then, step S140 is repeatedly executed until the maximum number of iterations is reached or the average distance threshold is less than a set value, and the iteration is exited.

[0056] In one embodiment, the calculation method of updating the original point set using the rotation matrix R and the translation matrix T is: i =Rp i +T. Where p' i is the updated original point.

[0057] S150, using the rotation matrix and translation matrix finally obtained after the iteration to calculate the rotation and translation values ​​of all scanning points to obtain transformation points, and confirming the wear information based on the transformation points and model points of the wear area.

[0058] In one embodiment, the final rotation matrix R and translation matrix T are used after the iteration to calculate the rotation and translation values ​​of all scanning points to obtain the transformation points. Each transformation point c' i The specific calculation method is as follows:

[0059] c' i =Rc i +T.

[0060] In one embodiment, confirming the wear information based on the transformation points and model points of the wear area includes: calculating the wear depth based on the transformation points and model points of the wear area, and confirming the wear area based on the intersection area area of ​​the transformation points and model points of the wear area. Specifically, intercepting point set 1 of the transformation points in area A, intercepting point set 2 of the model points in area A, calculating the maximum distance point in the vertical direction Z between the two point sets of point set 1 and point set 2, which is the maximum wear depth, calculating the intersection area area between the two point sets of point set 1 and point set 2, which is the wear area area, and finally displaying as shown below. Figure 5 shown.

[0061] In one embodiment, the Egien computing library is used for accelerated calculation, and KDtree is used for accelerated search. Through on-site modeling, calibration, and control of the number and quality of reasonable matching points, as well as setting of reasonable matching thresholds, the overall algorithm can detect a wear depth threshold of 0.2mm, and can achieve a basic stable operating speed of 0.05ms for single contour wear calculation on a CPU13700 hardware device, and can also support a real-time acquisition speed of 20KhZ, effectively improving the system operation speed and performance.

[0062] In the prior art, since the gradient of the rail at the inflection point is very large, the line scan camera 210 is easily distorted at the position with a large gradient, resulting in failure to scan the real inflection point, which causes the breakpoint of the scanning point in the prior art to be at a different position from the inflection point of the standard model. This causes a large error in the prior art when aligning in the coarse matching stage, which in turn causes an increase in the number of iterations and errors in the refined matching stage, which not only affects the calculation speed and accuracy, but may even cause the calculation to be stuck without convergence.

[0063] The track wear detection method of the present application generates a first model based on the scanning point set generated on site to achieve rapid on-site modeling, determines the original points involved in the matching in the scanning point set based on a preset matching range to achieve rapid on-site calibration, and iteratively calculates the rotation matrix and translation matrix between the original point and the target point through matching weights to achieve rapid iterative calculation. The track wear detection method of the present application allows on-site staff to perform rapid on-site modeling and calibration before the operation without the need for development experience, or to perform on-site track model reconstruction and calibration at regular fixed points, which can enhance anti-interference and facilitate maintenance. In addition, the embodiment of the present invention has been designed at the business level in a targeted manner, and the matching weights are set by introducing the prior knowledge of the on-site operation through the weight confirmation module, which effectively reduces the algorithm matching calculation complexity and the number of iterations, and reduces the interference of noise points, thereby improving the algorithm performance and greatly improving the algorithm operation speed, and can support faster scanning speeds, and can further support the scanning device to move along the track at high speed to scan and calculate the results at the same time. In addition, the method of the embodiment of the present invention first performs basic rough matching (calculating the rotation and translation matrices for the first time and performing rough alignment operations), and then performs refined matching (iterative rotation and translation matrices). During the rough matching process, the inflection point of the first model and the breakpoint of the scanning point are forced to be the same point, thereby calculating the rotation and translation matrices and performing rough alignment operations. This not only ensures the accuracy of the rough alignment, but also avoids the waste of computing power in the subsequent repeated iterations of refined matching.

[0064] [Example 2]

[0065] Figure 6is a schematic diagram of a rail wear detection device provided in the second embodiment of the present invention. The device 600 can be implemented by software and / or hardware, and can generally be integrated into a rail wear detection system such as Figure 6 As shown, the device 600 includes a data receiving module 610 , a model generating module 620 , a weight confirming module 630 , a parameter iteration module 640 and a wear calculating module 650 .

[0066] The data receiving module 610 is used to receive a set of scanning points of a track profile to be measured, where the track profile to be measured at least includes a worn area, a stable area, an untested area, and an auxiliary area.

[0067] In one embodiment, reference Figure 2 The scanning point set of the track profile to be measured is obtained by an image acquisition device 200, which includes a high-frequency laser line scanning camera 210 arranged at a 45° downward angle toward both sides of the track 220, and the height of the lens of the camera 210 from the track surface is h=25-35cm. The high-frequency laser is a laser in the infrared band of 606nm or 808nm, and the sampling frequency of the high-frequency laser is greater than or equal to 20KhZ.

[0068] In one embodiment, reference Figure 3 , the track profile to be measured includes a worn area A, a stable area B, an untested area C, an auxiliary area D and an interference area E. In the present embodiment, the worn area A includes the top surface of the track profile to be measured which is easily crushed or rubbed by the wheels. In the present embodiment, wear usually occurs in this area. Therefore, area A cannot participate in the matching, because if wear occurs, this area will not be consistent with the model, which will result in large iteration errors and time-consuming iterations. The stable area B is the top side that is close to the wheel hub but is not easily crushed or rubbed by the wheel hub. This area is not prone to wear and debris is not allowed to adhere. It is a relatively stable area. The present embodiment is mainly used for matching the model with the scanning point. The untested area C is Figure 2 The area that the middle camera 210 cannot detect due to being blocked by the side corner 211 does not participate in matching or wear assessment and is only used for display. The auxiliary area D is the part of the track profile that is used for support and will not wear. This area is similar to the B area and is relatively stable. This embodiment is used for auxiliary matching. The interference area E is the part of the bottom of the track profile that is blocked or buried by debris. This part does not participate in matching either.

[0069] The model generation module 620 is used to pre-process the scanning point set to generate a modeling point set, and to generate a first model based on the modeling point set, wherein the first model is used to generate model points.

[0070] In one embodiment, the model generation module 620 further includes a noise removal module, a contour point filling module, a breakpoint mapping module, and a collection point compensation module. The noise removal module is used to remove noise points of a frame of scanning point set, and the noise points at least include non-track points near the scanning breakpoints at the end of the stable area; the contour point filling module is used to fill the contour points of the uninspected area according to the standard track, and the breakpoint mapping module is used to force the scanning breakpoints to be mapped as model inflection points; the collection point compensation module is used to compensate the contour collection points of the auxiliary area according to the standard track; and the collection points of the wear area are compensated according to the top length of the standard track.

[0071] Specifically, after the image acquisition device 200 enters the scene, it first acquires a frame of contour point data, displays the acquired contour point data on a preset display interface, and manually removes the Figure 3 The points (noise) and non-rail points (noise) in the C area in the scan point are detected, and the end of the B area is detected at the same time to confirm that there are no noise points near the breakpoint 221. If there are noise points, the noise points are deleted to eliminate the problem that the camera scan is easily distorted in places with steep gradients. The C area points corresponding to the national standard track are filled in the blank C area of ​​the scanning point. This operation forces the breakpoint 221 of the scanning point to be mapped as the inflection point of the new model (first model). Referring to the tailing length of the E area of ​​the national standard track, the acquisition points are compensated or cut to eliminate the problem of occlusion or coverage on the site or the problem of excessive laser scanning on the site. The compensated acquisition points are used for modeling calibration. Referring to the top length of the A area under the half-track size of the national standard track, the acquisition points are compensated. In this embodiment, the above operations can be performed by pressing and clicking on the display interface preset by the front-end software. Therefore, the staff only needs simple operations to complete the contour modeling (generation of the first model) adapted to the site. Even if the camera acquisition state changes under different outdoor temperatures and humidity conditions, the best matching new model (first model) can still be achieved through on-site rapid modeling, thereby greatly improving the anti-interference ability of the system.

[0072] In this embodiment, as long as the points in area B and area D are stable and accurate, the first model can run stably. Although the country has a unified railway track size standard, the standard track point model can be directly used as a benchmark, but in order to improve system stability, maintenance convenience, and algorithm operation speed and accuracy, this embodiment designs a fast modeling module, which only uses some points in areas B and D in the national standard model, combined with the contour points collected on site, to quickly generate a first model adapted to the site.

[0073] The weight confirmation module 630 is used to determine the original points involved in the matching in the scanning point set according to the preset matching range, and set the matching weight based on the distance ranking between the original points and the model points.

[0074] In one embodiment, the weight confirmation module 630 includes a range setting module, which is used to select a first number of scanning points in a first length range of the stable area B adjacent to the unchecked area as the first original points participating in the matching, and select a second number of scanning points in a second length range of the auxiliary area D adjacent to the unchecked area C as the second original points participating in the matching. Confirm the index value of the inflection point (for example, the specific serial number of the inflection point after all scanning points are sorted) and enter it through the input window on the preset display interface. See also Figure 4 , B1 and D1 are the selection ranges of the original points participating in the matching in the B and D areas. In one embodiment, the range setting principle is that there should not be too many points participating in the iteration in the B area to prevent the points in the wear area A from entering and causing interference, and there should not be too many points participating in the matching in the D area to prevent the matching algorithm from over-optimizing the D area. In one embodiment, the number of points participating in the matching in the B1 and D1 areas can be kept basically the same. The on-site personnel can complete the calibration by clicking the mouse on the preset display interface to select the above matching range.

[0075] Specifically, in one embodiment, the B1 area can be 1 / 3-2 / 3 of the length of the B area, for example, it can be 1 / 2; when the number of points collected in the B1 area is less than a preset threshold, for example, 30, no downsampling is performed and all parameters are iterated; when the number of collected points in the B1 area is greater than the preset threshold, for example, 30, 30 points near the breakpoint (inflection point) 221 are completely retained (to prevent downsampling from destroying the accurate position of the inflection point), and the points at other remaining positions of B1 other than the 30 points near the breakpoint 221 are uniformly downsampled. The default parameter can be 6 times downsampling. During the deployment phase, the parameters can be adjusted through trial operation. If it is found that the number of participating points is too large, for example, greater than 45, the number of points can be reduced by increasing the downsampling multiple (for example, to 7-10 times). In actual tests, a total of about 45 iterative matching points in the B1 area have the best effect in terms of performance and speed.

[0076] Further, in the embodiment of the present invention, the length of the D1 region can be 1.5 times the length of the B1 region (because B1 is close to the camera and has a higher point density, in order to ensure that the number of candidate points in the two regions is basically the same, the point selection range of the D1 region is expanded). When the number of acquisitions in the D1 region is less than 30, no downsampling is performed, and all parameter iterations are performed; when the number of acquisition points in the D1 region is greater than the preset threshold, such as 30, 30 points near the breakpoint (inflection point) 221 are completely retained (to prevent downsampling from destroying the accurate position of the inflection point), and the points of the remaining positions of D1 other than the 30 points near the breakpoint 221 are uniformly downsampled. The default parameter can be 6 times downsampling. The parameters can be adjusted and set through trial operation during the deployment phase. If it is found that there are too many participating points, such as more than 45, the number of points can be reduced by increasing the downsampling multiple (for example, increasing it to 7-10 times). In actual tests, the performance and speed of iterative matching points in the D1 region are the best.

[0077] In one embodiment, the weight confirmation module 630 also includes a weight setting module, which is used to set the corresponding weight of the point with a smaller distance between the original point and the model point and ranked at the front by the first proportion threshold as the first weight, and set the corresponding weight of the point with a larger distance between the original point and the model point and ranked at the back by the second proportion threshold as the second weight. In one embodiment, the first proportion threshold may be 60-80%, the second proportion threshold may be 10-30%, the first weight may be 1, and the second weight may be 0. For example, the corresponding weight of the points with a smaller distance between the original point and the model point and ranked at the front by 80% may be set to 1, and the corresponding weight of the points with a larger distance between the original point and the model point and ranked at the back by 20% may be set to 0.

[0078] Specifically, in one embodiment, all scanning points are defined as: The original points involved in the matching in the scan points are defined as: The nearest model point is defined as: The distance between the original point and the model point is calculated as:

[0079] In one embodiment, the distance d of all original points participating in the matching (assuming N points) is calculated and sorted from large to small, and the index value is The distance value is marked as d thresh , the weight function is:

[0080]

[0081] The parameter iteration module 640 is used to identify the target point closest to each original point in the model points, and iteratively calculate the rotation matrix and translation matrix between the original point and the target point based on the matching weight.

[0082] In one embodiment, the parameter iteration module 640 searches for the points closest to the original points in the model points and combines them into a new point set, which is collectively referred to as the target point below. The calculation formula for the nearest point (Euclidean distance) is:

[0083]

[0084] In one embodiment, the parameter iteration module 640 includes a weighted center calculation module, which is used to calculate the weighted centers of the original point set and the target point set based on the matching weights, respectively, the original point set is a set of all original points, and the target point set is a set of all target points;

[0085] Specifically, the weighted centers of the original point set and the target point set are calculated respectively. In this embodiment, the weighted center of the original point set is calculated as follows: The weighted center of the target point set is calculated as follows:

[0086] In one embodiment, the parameter iteration module 640 further includes a point set generation module, which calculates a first point set after the original point set is decentralized and a second point set after the target point set is decentralized based on the weighted center.

[0087] Among them, the calculation method of the first point set after decentralization is: The calculation method of the second point set after decentralization is:

[0088] In one embodiment, the parameter iteration module 640 further includes a variance calculation module, which is used to calculate the weighted covariance matrix of the first point set and the second point set;

[0089] Specifically, the first point set is defined as: The second point set is defined as: The weight set is defined as: w n×N =diag(w0,w1,w2,w3...wn). The calculation method of the rotation matrix is:

[0090] In one embodiment, the parameter iteration module 640 further includes a rotation calculation module for calculating a rotation matrix by decomposing the covariance matrix through SVD. The covariance matrix decomposition method is: S = UDV T ; The calculation method of the rotation matrix R is:

[0091] In one embodiment, the parameter iteration module 640 further includes a translation calculation module, configured to calculate a translation matrix T using weighted centers of the first point set and the second point set.

[0092] In one embodiment, the translation matrix T is calculated as follows: In this embodiment, the weighted centers of the first point set and the second point set are used for calculation, and calculation cannot be performed through inflection points or breakpoints.

[0093] In one embodiment, the parameter iteration module 640 further includes a point set updating module for updating the original point set using the rotation matrix R and the translation matrix T. The parameter iteration module 640 further includes an iterative calculation module for repeatedly executing the steps after determining the original points involved in the matching in the scanning point set according to the preset matching range in the first embodiment until the maximum number of iterations is reached or the average distance threshold is less than the set value.

[0094] Specifically, the original point set is updated with the rotation matrix R and the translation matrix T Then, the functions of each module of the parameter iteration module 640 are repeatedly executed until the maximum number of iterations is reached or the average distance threshold is less than the set value, and the iteration is exited.

[0095] In one embodiment, the calculation method of updating the original point set using the rotation matrix R and the translation matrix T is: i =Rp i +T. Where p' i is the updated original point.

[0096] The wear calculation module 650 is used to calculate the rotation and translation values ​​of all scanning points using the rotation matrix and translation matrix finally obtained after the iteration to obtain the transformation points, and confirm the wear information based on the transformation points and model points of the wear area.

[0097] In one embodiment, the final rotation matrix R and translation matrix T after the iteration are used to calculate the rotation and translation values ​​of all scanning points to obtain the transformation points. The specific calculation method of each transformation point c_i^, is as follows:

[0098] c_i^,=Rc_i+T.

[0099] In one embodiment, confirming the wear information based on the transformation points and model points of the wear area includes: calculating the wear depth based on the transformation points and model points of the wear area, and confirming the wear area based on the intersection area area of ​​the transformation points and model points of the wear area. Specifically, intercepting point set 1 of the transformation points in area A, intercepting point set 2 of the model points in area A, calculating the maximum distance point in the vertical direction Z between the two point sets of point set 1 and point set 2, which is the maximum wear depth, calculating the intersection area area between the two point sets of point set 1 and point set 2, which is the wear area area, and finally displaying as shown below. Figure 5 shown.

[0100] In one embodiment, the Egien computing library is used for accelerated calculation, and KDtree is used for accelerated search. Through on-site modeling, calibration, and control of the number and quality of reasonable matching points, as well as setting of reasonable matching thresholds, the overall algorithm can detect a wear depth threshold of 0.2mm, and can achieve a basic stable operating speed of 0.05ms for single contour wear calculation on a CPU13700 hardware device, and can also support a real-time acquisition speed of 20KhZ, effectively improving the system operation speed and performance.

[0101] The above-mentioned rail wear detection device can execute the rail wear detection method provided in any embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not fully described in this embodiment, please refer to the rail wear detection method provided in any embodiment of the present invention. Since the rail wear detection device introduced above is a device that can execute the rail wear detection method in the embodiment of the present invention, based on the rail wear detection method introduced in the embodiment of the present invention, the technical personnel in this field can understand the specific implementation mode of the rail wear detection device of this embodiment and its various variations, so how the rail wear detection device implements the rail wear detection method in the embodiment of the present invention will not be described in detail here. As long as the technical personnel in this field implement the device used in the rail wear detection method in the embodiment of the present invention, it belongs to the scope of protection of this application.

[0102] [Example 3]

[0103] Figure 7 FIG. 2 shows a schematic diagram of the structure of a rail wear detection system provided by Embodiment 3 of the present invention. Figure 7 As shown, the rail wear detection system 700 may include an image acquisition device 200 and at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. Various programs and data required for the operation of the rail wear detection system 700 can also be stored in the RAM 13. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0104] Multiple components in the rail wear detection system 700 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the rail wear detection system 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0105] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a single-chip microcomputer, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a track wear detection method.

[0106] In some embodiments, the rail wear detection method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the rail wear detection system 700 of the rail wear detection system 700 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the rail wear detection method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the rail wear detection method in any other suitable manner (e.g., by means of firmware).

[0107] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0108] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0109] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0110] To provide interaction with a user, the systems and techniques described herein may be implemented on a mobile terminal having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the mobile terminal. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0111] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0112] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0113] [Example 4]

[0114] Embodiment 4 of the present invention further provides a computer storage medium storing a computer program, wherein the computer program, when executed by a computer processor, is used to execute the rail wear detection method described in any of the above embodiments of the present invention.

[0115] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM, or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0116] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0117] The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, radio frequency (RF), etc., or any suitable combination of the foregoing.

[0118] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A rail wear detection method, characterized in that: include: receiving a set of scanning points of a track profile to be measured, wherein the track profile to be measured includes at least a worn area, a stable area, an untested area, and an auxiliary area; Preprocessing the scanning point set to generate a modeling point set, generating a first model based on the modeling point set, wherein the first model is used to generate model points; Determine the original points involved in the matching in the scan point set according to the preset matching range, and set the matching weight based on the distance sorting between the original points and the model points; Confirm the target point closest to each original point in the model points, and iteratively calculate the rotation matrix and translation matrix between the original point and the target point based on the matching weight; The rotation matrix and translation matrix finally obtained after the iteration are used to calculate the rotation and translation values ​​of all scanning points to obtain the transformation points, and the wear information is confirmed based on the transformation points and model points of the wear area.

2. The rail wear detection method according to claim 1, characterized in that: Preprocessing the scanning point set to generate a modeling point set includes: Removing noise points of a frame of scan point set, wherein the noise points at least include non-track points near the scanning breakpoints at the end of the stable region; Filling the uninspected area with contour points according to the standard track, and forcibly mapping the scanning breakpoints to model inflection points; Compensating the contour acquisition points of the auxiliary area according to the standard track; The acquisition points of the worn area are compensated according to the top length of the standard track.

3. The rail wear detection method according to claim 1, characterized in that: Determining the original points involved in the matching in the scanning point set according to the preset matching range includes: Select a first number of scanning points as the first original points involved in matching in a first length range of the stable area adjacent to the undetected area, A second number of scanning points are selected in a second length range of the auxiliary area adjacent to the uninspected area as second original points involved in matching.

4. The rail wear detection method according to claim 1, characterized in that: The setting of matching weights based on the distance order between the original points and the model points comprises: The weight corresponding to the point with the smallest distance between the original point and the model point and the first proportion threshold is set as the first weight, The weight corresponding to the point whose distance between the original point and the model point is larger and which is ranked closer to the second proportion threshold is set as the second weight.

5. The rail wear detection method according to claim 4, characterized in that: The iterative calculation of the rotation matrix and translation matrix between the original point and the target point based on the matching weights comprises: Based on the matching weights, weighted centers of an original point set and a target point set are calculated respectively, wherein the original point set is a set of all original points, and the target point set is a set of all target points; Calculate a first point set after the original point set is decentralized based on the weighted center, and a second point set after the target point set is decentralized; Calculating a weighted covariance matrix of the first point set and the second point set; SVD decomposes the covariance matrix and calculates the rotation matrix; Calculate the translation matrix T using the weighted centers of the first point set and the second point set; The original point set is updated using the rotation matrix R and the translation matrix T, and the steps after determining the original points involved in the matching in the scan point set according to the preset matching range are repeated until the maximum number of iterations is reached or the average distance threshold is less than the set value.

6. The rail wear detection method according to any one of claims 1 to 5, characterized in that: The confirming of wear information based on the transformation points and model points of the wear area includes: The wear depth is calculated based on the transformed points and the model points of the wear area, and the wear area is determined based on the intersection area of ​​the transformed points and the model points of the wear area.

7. The rail wear detection method according to claim 6, characterized in that: The calculation of the wear depth based on the transformation points and the model points of the wear area, and the confirmation of the wear area based on the intersection area of ​​the transformation points and the model points of the wear area include: Intercept the point set 1 of the transformation points in the wear area, and intercept the point set 2 of the model points in the wear area. Calculate the maximum vertical distance point between two point sets, which is the maximum wear depth. Calculate the intersection area between the two point sets, which is the wear area.

8. A rail wear detection device, characterized in that: include: A data receiving module, used for receiving a scanning point set of a track profile to be measured, wherein the track profile to be measured includes at least a worn area, a stable area, an untested area and an auxiliary area; A model generation module, used for preprocessing the scanning point set to generate a modeling point set, and generating a first model based on the modeling point set, wherein the first model is used to generate model points; A weight confirmation module, used to determine the original points involved in the matching in the scanning point set according to a preset matching range, and set the matching weight based on the distance ranking between the original points and the model points; A parameter iteration module, used for confirming the target point closest to each original point in the model points, and iteratively calculating the rotation matrix and translation matrix between the original point and the target point based on the matching weight; The wear calculation module is used to calculate the rotation and translation values ​​of all scanning points using the rotation matrix and translation matrix finally obtained after the iteration to obtain the transformation points, and confirm the wear information based on the transformation points and model points of the wear area.

9. A rail wear detection system, characterized in that: include: An image acquisition device, used to generate a scanning point set for scanning the track; as well as one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the track wear detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the track wear detection method as described in any one of claims 1 to 7 is implemented.

Images