Method and system for detecting alternate side abrasion of steel rail

Abstract

The invention discloses a method and system for detecting alternate side grinding of a steel rail, and the method comprises the steps: obtaining a three-dimensional profile of a steel rail profile in an abnormal shaking region of a subway, and generating an alternate side grinding waveform at a side grinding position according to the three-dimensional profile of the steel rail; and according to the alternate side grinding waveform, the occurrence position of the alternate side grinding phenomenon of the steel rail is determined through the side grinding recognition model. The method has remarkable significance in quantitative evaluation of the steel rail alternate side abrasion phenomenon, the measurement accuracy is improved, and rapid and efficient recognition of steel rail alternate side abrasion is achieved through the side abrasion recognition model. In addition, important technical support can be provided for subway operation maintenance and safety management.

Description

Technical Field

[0001] The present invention relates to the field of rail transportation technology, and in particular to a method and system for detecting alternate side wear of rails. Background Art

[0002] In recent years, the problem of low-frequency subway swaying has become increasingly common. During field surveys of swaying sections on multiple lines, alternating rail side wear was found. Alternating rail side wear usually occurs on straight lines or gentle curves with a large radius. Specifically, the rail light band changes periodically between the top and the inside of the rail. The wear positions of the left and right rails are exactly opposite, and the wear positions on the inside appear alternately. Studies have found that alternating subway side wear will cause periodic changes in the wheel-rail contact relationship, causing periodic disturbances to the vehicle, resulting in low-frequency abnormal swaying of the subway vehicle, which has become a prominent problem affecting the comfort of subway passengers and the safety of operation.

[0003] Rail alternating side grinding is similar to rail corrugation, which is a special form of abnormal wear. The surface of the working edge of the rail head shows periodic unevenness. There is a big difference between rail alternating side grinding and traditional corrugation. The wavelength of rail alternating side grinding is much longer than that of traditional corrugation, up to about 12 meters, but the amplitude of unevenness is small, usually about 1 mm. Due to its long wavelength and small amplitude, there is still a lack of direct measurement methods for the waveform of rail alternating side grinding. In actual engineering, the wavelength of alternating side grinding can only be roughly judged by visually observing the light band on the rail surface.

[0004] However, when the rail light band is not obvious or operations such as grinding damage the light band, the problem of alternating side grinding of the rail cannot be effectively detected, causing trouble for subway engineering operations and maintenance. Summary of the invention

[0005] The purpose of the present invention is to provide a solution for detecting alternate side wear of rails, so as to solve the existing problem that alternate side wear of rails cannot be effectively detected.

[0006] In order to solve the above technical problems, an embodiment of the present invention provides a method for detecting alternating side wearing of rails, the method comprising: obtaining a three-dimensional profile of the rail profile in an abnormal swaying area of ​​a subway, and generating an alternating side wearing waveform at the side wearing position according to the three-dimensional profile of the rail; constructing a side wearing recognition model according to the alternating side wearing waveform; collecting the three-dimensional profile of the rails at different positions to generate corresponding alternating side wearing waveforms, and determining the occurrence location of the alternating side wearing phenomenon of the rails using the side wearing recognition model.

[0007] Preferably, the three-dimensional profile of the rail is obtained using the following steps: using preset distance intervals, collecting the three-dimensional profiles of the rails in different longitudinal track sections in the abnormal swaying area of ​​the subway; longitudinally connecting the three-dimensional profiles of the rails in different longitudinal track sections to obtain the three-dimensional profile of the rail, wherein the transverse direction of the three-dimensional profile of the rail represents the horizontal direction perpendicular to the rail, the longitudinal direction represents the extension direction of the rail, and the vertical direction represents the plumb line direction of the rail.

[0008] Preferably, the step of generating an alternating side grinding waveform at the side grinding position according to the three-dimensional profile of the rail includes: converting the three-dimensional profile of the rail into an original height cloud map; determining the horizontal coordinate at the maximum value of the profile height difference by identifying the highest peak and the lowest trough in the original height cloud map; intercepting the height cloud map within a preset range and slicing the intercepted height cloud map to form a curve of vertical height changing with the longitudinal direction under different horizontal coordinate conditions, which is recorded as an alternative waveform curve; and obtaining an alternating side grinding waveform according to the horizontal coordinate at the maximum value of the profile height difference and multiple alternative waveform curves.

[0009] Preferably, the step of intercepting the height cloud map within a preset range and aligning and slicing it to form a curve of vertical height changing with the longitudinal direction under different horizontal coordinate conditions, which is recorded as an alternative waveform curve, includes: generating an allowable value of the horizontal coordinate based on the horizontal coordinate at the maximum value of the profile height difference; intercepting the height cloud map within the range between the allowable value of the horizontal coordinate and the horizontal coordinate on the inner side of the track in the original height cloud map; and slicing the intercepted height cloud map along the longitudinal direction at preset intervals to obtain multiple alternative waveform curves.

[0010] Preferably, the step of determining the horizontal coordinate of the maximum value of the profile height difference by identifying the highest peak and the lowest trough in the original height cloud map includes: determining the longitudinal position range between the longitudinal position of the highest peak and the longitudinal position of the lowest trough according to the highest peak and the lowest trough in the original height cloud map; analyzing the variation characteristics of the vertical height with the lateral direction within the longitudinal position range to determine the horizontal coordinate of the maximum value of the profile height difference.

[0011] Preferably, the step of obtaining the alternating side grinding waveform according to the horizontal coordinate at the maximum value of the profile height difference and multiple alternative waveform curves includes: based on the multiple alternative waveform curves, extracting the curve of the vertical height changing with the longitudinal direction under the horizontal coordinate condition of the maximum value of the profile height difference, and recording it as the target waveform curve; de-averaging the target waveform curve to obtain the alternating side grinding waveform.

[0012] Preferably, the alternating side grinding waveform includes: an alternating side grinding waveform corresponding to the left rail and an alternating side grinding waveform corresponding to the right rail.

[0013] Preferably, the side wear recognition model is constructed according to the alternating side wear waveform, which includes: obtaining the image to be input according to the alternating side wear waveform corresponding to the left rail and the alternating side wear waveform corresponding to the right rail in combination with an image processing method; obtaining input images corresponding to different levels according to the image to be input in combination with a hierarchical feature processing method; shifting the input image by a preset shift window, intercepting the image corresponding to each shift window and performing feature analysis on the intercepted image to obtain feature information corresponding to different levels; and constructing the side wear recognition model using the feature information corresponding to all levels and the position of the rails as training data.

[0014] Preferably, the levels include: a low-level level, an intermediate level and a high-level level; wherein the feature information corresponding to the low-level level includes but is not limited to: edge features; the feature information corresponding to the intermediate level includes but is not limited to: shape features and structural features; the feature information corresponding to the high-level level includes but is not limited to: alternating side grinding prediction result features.

[0015] On the other hand, an embodiment of the present invention also provides a system for detecting alternating side grinding of rails, the system comprising: a characteristic waveform generation module, configured to obtain a three-dimensional profile of the rail profile in an abnormal subway swaying area, and generate an alternating side grinding waveform at the side grinding position based on the three-dimensional profile of the rail; an identification model construction module, configured to construct a side grinding identification model based on the alternating side grinding waveform; a real-time identification module, configured to collect the three-dimensional profile of the rail to be tested at different positions and generate corresponding alternating side grinding waveforms, and use the side grinding identification model to determine the occurrence location of the alternating side grinding of the rail.

[0016] Compared with the prior art, one or more embodiments of the above scheme may have the following advantages or beneficial effects:

[0017] The present invention proposes a method and system for detecting alternate side wear of rails. The method and system obtain the rail profile of the abnormal swaying area of ​​the subway, and use the rail profile to obtain a three-dimensional profile, which can more comprehensively capture the subtle changes on the rail surface and provide an accurate basis for the analysis of alternate side wear of rails; and after obtaining the three-dimensional profile, the alternating side wear waveform is obtained, which can provide an intuitive basis for analyzing the degree and distribution of alternate side wear of rails; finally, the side wear recognition model established can automatically and intelligently determine the occurrence location of the alternate side wear of rails. The present invention realizes the rapid and efficient identification and evaluation of alternate side wear of rails through the application of the side wear recognition model, which is of great significance for quantitatively evaluating the degree of alternate side wear of rails. Furthermore, it is expected to provide important technical support for subway operation maintenance and safety management.

[0018] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0020] Figure 1 Schematic diagram of the steps of a method for detecting alternate side wear of rails according to an embodiment of the present application.

[0021] Figure 2 It is a schematic diagram of the process of obtaining the alternating side wear waveform in the method for detecting the alternating side wear of rails in an embodiment of the present application.

[0022] Figure 3 It is a schematic diagram of a scenario of alternating side wear of rails of a first example track segment in the method for detecting alternating side wear of rails according to an embodiment of the present application.

[0023] Figure 4 It is a schematic diagram of a scenario of alternating side wear of rails of a second example track segment in the method for detecting alternating side wear of rails according to an embodiment of the present application.

[0024] Figure 5 This is an example diagram of some results of the three-dimensional profile of the rail in the method for detecting alternating side wear of the rail according to an embodiment of the present application.

[0025] Figure 6 This is an example diagram of the results of the original height cloud map in the method for detecting alternating side wear of rails according to an embodiment of the present application.

[0026] Figure 7 This is an example diagram of the results of multiple alternative waveform curves in the method for detecting alternate side wear of rails according to an embodiment of the present application.

[0027] Figure 8 This is an example diagram of the results of the profile height difference in the method for detecting alternating side wear of rails according to an embodiment of the present application.

[0028] Fig. 9 This is an example diagram of the result of the alternating side wear waveform in the method for detecting alternating side wear of rails in an embodiment of the present application.

[0029] Fig.10 It is a schematic diagram of the process of constructing a side wear recognition model in the method for detecting alternating side wear of rails according to an embodiment of the present application.

[0030] Fig.11 Schematic diagram of the structure of a system for detecting alternate side wear of rails according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] The following will describe the implementation methods of the present invention in detail with reference to the accompanying drawings and embodiments, so that the implementation process of how the present invention applies technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly. It should be noted that as long as there is no conflict, the various embodiments of the present invention and the various features in the embodiments can be combined with each other, and the technical solutions formed are all within the protection scope of the present invention.

[0032] In addition, the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions. Also, although a logical sequence is shown in the flowchart, in some cases, the steps shown or described can be performed in a sequence different from that here.

[0033] The terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments. Unless the context clearly indicates otherwise, the singular forms "one", "one" and "item" used herein are also intended to include plural numbers. It should also be understood that the terms "include" and / or "comprise" used herein specify the existence of stated features, integers, steps, operations, units and / or components, without excluding the existence or addition of one or more other features, integers, steps, operations, units, components and / or combinations thereof.

[0034] In recent years, the problem of low-frequency subway swaying has become increasingly common. During field surveys of swaying sections on multiple lines, alternating rail side wear was found. Alternating rail side wear usually occurs on straight lines or gentle curves with a large radius. Specifically, the rail light band changes periodically between the top and the inside of the rail. The wear positions of the left and right rails are exactly opposite, and the wear positions on the inside appear alternately. Studies have found that alternating subway side wear will cause periodic changes in the wheel-rail contact relationship, causing periodic disturbances to the vehicle, resulting in low-frequency abnormal swaying of the subway vehicle, which has become a prominent problem affecting the comfort of subway passengers and the safety of operation.

[0035] Rail alternating side grinding is similar to rail corrugation, which is a special form of abnormal wear. The surface of the working edge of the rail head shows periodic unevenness. There is a big difference between rail alternating side grinding and traditional corrugation. The wavelength of rail alternating side grinding is much longer than that of traditional corrugation, up to about 12 meters, but the amplitude of unevenness is small, usually about 1 mm. Due to its long wavelength and small amplitude, there is still a lack of direct measurement methods for the waveform of rail alternating side grinding. In actual engineering, the wavelength of alternating side grinding can only be roughly judged by visually observing the light band on the rail surface.

[0036] However, when the rail light band is not obvious or operations such as grinding damage the light band, the problem of alternating side grinding of the rail cannot be effectively detected, causing trouble for subway engineering operations and maintenance.

[0037] Figure 1 This is a schematic diagram of the steps of the method for detecting alternate side wear of rails according to an embodiment of the present application. Figure 1 , the specific steps and procedures of the method for detecting alternate side wear of rails (hereinafter referred to as “alternate side wear detection method of rails”) according to an embodiment of the present invention are described.

[0038] Step S110, obtaining a three-dimensional profile of a rail profile in an abnormally shaking area of ​​a subway, and generating an alternating side grinding waveform at a side grinding position according to the three-dimensional profile of the rail profile.

[0039] In this embodiment, the measurement problem of long wavelength and low amplitude longitudinal unevenness of the rail surface is transformed into the measurement problem of rail profile. Therefore, the rail profile of the subway abnormal swaying area is obtained by using a mounted track inspection device, and then the three-dimensional profile of the rail profile is obtained according to the rail profile of the subway abnormal swaying area. Finally, the alternating side grinding waveforms at different side grinding positions are generated according to the three-dimensional profile of the rail profile, so as to construct a side grinding recognition model.

[0040] In this embodiment, there is no limit to the area of ​​abnormal subway swaying, and a reasonable selection can be made according to actual application requirements.

[0041] Optionally, in order to ensure the smoothness of the alternating side grinding waveform, the alternating side grinding waveform may preferably include: an alternating side grinding waveform corresponding to the left rail and an alternating side grinding waveform corresponding to the right rail. Therefore, a set of left rail and right rail profiles may be tested above each soft sleeper, and a three-dimensional profile of the left rail profile and a three-dimensional profile of the right rail profile may be formed respectively, and finally the alternating side grinding waveform corresponding to the left rail and the alternating side grinding waveform corresponding to the right rail may be obtained.

[0042] In one embodiment, in the step of obtaining the three-dimensional profile of the rail profile in the abnormal swaying area of ​​the subway, first, the three-dimensional profiles of the rails of different longitudinal tracks are collected, and then the three-dimensional profiles of the rails of different longitudinal tracks are processed in series longitudinally to obtain the three-dimensional profile of the rail profile. Specifically, step S110 includes the following sub-steps A1-A2:

[0043] Sub-step A1, using a preset distance interval, collects the three-dimensional profiles of the rails of different longitudinal track sections in the subway abnormal swaying area.

[0044] Sub-step A2, longitudinally connecting the three-dimensional profiles of the rails of different longitudinal track sections in series to obtain the three-dimensional profile of the rails.

[0045] Among them, in sub-step A1, the specific value of the preset distance interval is not limited and can be reasonably selected according to actual application requirements.

[0046] Among them, in sub-step A2, the transverse direction of the three-dimensional contour of the rail profile represents the horizontal direction perpendicular to the rail, the longitudinal direction represents the extension direction of the rail, and the vertical direction represents the plumb line direction of the rail (ie, the vertical direction).

[0047] In one embodiment, in order to make the alternating side grinding waveform formed according to the three-dimensional profile more accurately reflect the alternating side grinding characteristics, multiple candidate waveform curves can be obtained, and then the final alternating side grinding waveform is obtained according to the multiple candidate waveform curves. Specifically, step S110 includes the following sub-steps B1-B4:

[0048] Sub-step B1, converting the three-dimensional profile of the rail into an original height cloud map.

[0049] Sub-step B2, determining the horizontal coordinate of the maximum value of the profile height difference by identifying the highest peak and the lowest trough in the original height cloud map.

[0050] Sub-step B3, intercepting the height cloud map within a preset range and slicing the intercepted height cloud map to form a curve of vertical height changing with the longitudinal direction under different horizontal coordinate conditions, which is recorded as an alternative waveform curve.

[0051] Sub-step B4, obtaining an alternating side grinding waveform according to the horizontal coordinate at the maximum value of the profile height difference and a plurality of alternative waveform curves.

[0052] Specifically, refer to Figure 2 First, the three-dimensional profile of the rail is converted into an original height cloud map, and then the highest peak and the lowest trough in the original height cloud map are identified to obtain the horizontal coordinate at the maximum value of the height difference of the profile. After obtaining the horizontal coordinate at the maximum value of the height difference of the profile, the height cloud map within a preset range is intercepted. After intercepting the height cloud map, the height cloud map is sliced ​​to obtain multiple alternative waveform curves. Finally, the waveform curve is determined according to the horizontal coordinate at the maximum value of the height difference of the profile and the multiple alternative waveform curves, and the waveform curve is used as the alternating side grinding curve.

[0053] Among them, in sub-step B1, the lateral direction in the original height cloud map represents the extension direction of the rail, the vertical direction represents the lateral two-phase direction of the rails on both sides, and the different color blocks in the original height cloud map represent the vertical height of the rail.

[0054] Among them, in sub-step B2, the side grinding area is determined according to the original height cloud map, and then the highest peak and the lowest trough are determined according to the side grinding area. The side grinding area represents the area corresponding to the original height cloud map when the vertical height of the rail is greater than the preset height threshold. The value of the preset height threshold is preferably 0.4 mm.

[0055] Furthermore, in sub-step B2, in the process of determining the horizontal coordinate at the maximum value of the height difference of the profile, the longitudinal position corresponding to the highest wave crest and the longitudinal position corresponding to the lowest wave trough are determined, and then, the change characteristics of the vertical height between the longitudinal positions along the transverse direction are analyzed to determine the horizontal coordinate at the position with the maximum height difference of the profile. Specifically, sub-step B2 includes the following steps B21-B22:

[0056] Step B21, determining the longitudinal position range between the longitudinal position of the highest peak and the longitudinal position of the lowest trough according to the highest peak and the lowest trough in the original height cloud map.

[0057] Step B22, analyzing the variation characteristics of the vertical height along the transverse direction within the longitudinal position range, and determining the horizontal coordinate at the maximum value of the profile height difference.

[0058] Specifically, the corresponding rail profile is determined according to the highest wave crest, and the corresponding rail profile is determined according to the lowest wave trough; after determining the corresponding rail profiles, the corresponding longitudinal positions are determined respectively, and then the longitudinal position range is obtained; then, according to the longitudinal position range and the corresponding rail profile, the variation characteristics of the vertical height with the transverse direction within the longitudinal position range are analyzed, and the abscissa at the point where the vertical height difference is the largest is selected as the abscissa at the point where the profile height difference is the largest.

[0059] Optionally, when calculating the height difference of the profile, the non-working edge is calculated first, and then the vertical height difference of the rail profile is calculated.

[0060] In sub-step B3, the value of the preset range is not limited and can be reasonably selected according to actual application requirements, for example, it can be 0-35 mm.

[0061] Wherein, in sub-step B3, the vertical height of the rail refers to the height corresponding to the direction of the plumb line at the position of the rail.

[0062] Further, in sub-step B3, in the process of obtaining the candidate waveform curves, firstly, according to the preset range, the horizontal coordinate allowable value that can be sliced ​​is obtained, then the height cloud map is intercepted from the original height cloud map according to the horizontal coordinate allowable value, and finally the intercepted height cloud map is sliced ​​along the longitudinal direction to obtain multiple candidate waveform curves. Specifically, sub-step B3 includes the following steps B31-B33:

[0063] Step B31, generating an allowable value of the horizontal coordinate based on the horizontal coordinate at the maximum value of the height difference of the profile formed above.

[0064] Step B32, intercepting the height cloud map within the range between the horizontal coordinate allowable value and the horizontal coordinate inside the track in the original height cloud map.

[0065] Step B33, slicing the intercepted height cloud image along the longitudinal direction according to preset intervals to obtain multiple candidate waveform curves.

[0066] Among them, in step B31, the horizontal coordinate allowable value represents an alternative horizontal coordinate value that can slice the intercepted height cloud image along the longitudinal direction. In an embodiment of the present invention, the horizontal coordinate allowable value is slightly larger than the horizontal coordinate at the maximum value of the above-mentioned profile height difference. The horizontal coordinate range formed between the horizontal coordinate allowable value and the horizontal coordinate inside the track is the above-mentioned preset range.

[0067] For example, when the horizontal coordinate at the maximum value of the profile height difference is 31 mm, the horizontal coordinate allowable value can be set to 35 mm, so as to form a preset range of 0-35 mm.

[0068] In step B32, the actual value of the track inner horizontal coordinate is not limited and can be reasonably selected according to actual application requirements. In the embodiment of the present invention, the track inner horizontal coordinate is preferably 0.

[0069] Among them, in step B33, according to the preset range and the preset interval, the abscissa for the final slicing process is obtained, and then the height cloud map is sliced ​​using the abscissa to form a curve of the vertical height changing with the longitudinal direction under the corresponding abscissa condition, which is recorded as the alternative waveform curve. The specific value of the preset interval is not limited and can be reasonably selected according to actual application requirements. For example, the preset interval can be 5mm.

[0070] Further, in sub-step B4, since the waveforms of different alternative waveform curves are similar, but the wave depth amplitudes are significantly different, in order to accurately describe the wave depth of the alternate side wear of the rail, the alternate side wear waveform is obtained according to the horizontal coordinate at the maximum value of the profile height difference and the alternative waveform curve. Specifically, sub-step B4 includes the following steps B41-B42:

[0071] Step B41, based on multiple candidate waveform curves, extract the curve of vertical height variation with longitudinal direction under the condition of the horizontal coordinate of the maximum value of the profile height difference, and record it as the target waveform curve.

[0072] Step B42, de-averaging the target waveform curve to obtain an alternating side grinding waveform.

[0073] Among them, in step B42, the de-averaging process is to subtract the mean of the corresponding dimension from the data of different dimensions in the obtained target waveform curve, so that each dimension of the input data in the alternating side grinding waveform is centered to 0.

[0074] Step S120: constructing a side wear recognition model according to the alternating side wear waveform.

[0075] After the alternating side wear waveform is obtained, the alternating side wear waveform is used as a training sample to construct a side wear recognition model, so that the occurrence position of the alternating side wear of the rail can be determined by using the side wear recognition model.

[0076] In one embodiment, the Swin Transformer algorithm is used to construct a side grind recognition model. First, the alternating side grind waveform is subjected to image processing such as image segmentation to obtain an input image; then, the input image is subjected to hierarchical feature processing to obtain input images of different resolutions, and the images of different resolutions are used as input images; then, feature information of the input image is extracted to construct a side grind recognition model. Specifically, step S120 includes the following sub-steps C1-C4:

[0077] Sub-step C1, obtaining an image to be input according to the alternating side grinding waveform corresponding to the left rail and the alternating side grinding waveform corresponding to the right rail in combination with an image processing method.

[0078] Sub-step C2, according to the image to be input, combined with the hierarchical feature processing method, to obtain input images corresponding to different levels.

[0079] Sub-step C3, using a preset shift window, shifting the input image, intercepting the image corresponding to each shift window and performing feature analysis on the intercepted image to obtain feature information corresponding to different levels.

[0080] Sub-step C4, using the feature information corresponding to all levels and the position of the rail as training data to build a side wear recognition model.

[0081] Specifically, the construction process of the side wear recognition model is divided into image segmentation and linear embedding, hierarchical feature construction, shift window strategy and position embedding.

[0082] The alternating side grinding waveform image corresponding to the left track is obtained according to the alternating side grinding waveform corresponding to the left track, and the alternating side grinding waveform image corresponding to the right track is obtained according to the alternating side grinding waveform corresponding to the right track. The alternating side grinding waveform image is divided into non-overlapping image blocks of fixed size to facilitate subsequent feature extraction and recognition. Each image block is flattened into a vector, and the flattened vector is mapped to a feature space of fixed dimension through a linear transformation layer. These feature vectors are used as the input image.

[0083] Combined with the hierarchical feature processing method, the hierarchical features are constructed by continuously reducing the resolution of the input image. Through the hierarchical features, the side wear recognition model can better capture the local and global information of the rail alternating side wear and improve the accuracy of the rail alternating side wear detection and recognition.

[0084] The shift window strategy is adopted to make windows of different layers overlap, so as to capture more characteristic information such as shape, space, frequency, etc. of alternating side grinding of rails and improve the robustness of detection.

[0085] By embedding the position of the rails, the feature information corresponding to all levels and the position of the rails are used as training data to ensure that the model can understand the spatial relationship of the alternating side wear image blocks and improve the accuracy of detection.

[0086] Further, in sub-step C3, the levels include: a low level, a middle level and a high level.

[0087] The feature information corresponding to the low-level layer includes but is not limited to edge features. The feature information corresponding to the intermediate layer includes but is not limited to shape features and structural features. The feature information corresponding to the high-level layer includes but is not limited to alternating side grinding prediction result features.

[0088] Step S130, collecting the three-dimensional profiles of the rail to be tested at different positions and generating corresponding alternating side wear waveforms, and using the side wear recognition model to determine the location where the alternating side wear phenomenon of the rail occurs.

[0089] After obtaining the side wear recognition model, when determining the occurrence location of the alternating side wear phenomenon of the rail to be tested, the three-dimensional profile at different positions of the rail to be tested can be collected in a similar manner to the above-mentioned step S110, and the corresponding alternating side wear waveform can be generated. Finally, based on the alternating side wear waveform and the constructed side wear recognition model, the occurrence location of the alternating side wear phenomenon of the rail can be determined.

[0090] Example

[0091] Based on the above embodiment, the specific process of applying the method for detecting alternate side wear of rails according to the embodiment of the present invention to a certain rail at a certain location is described below.

[0092] When alternating side wear occurs, the rail light band changes periodically between the rail top and the inner side. The wear positions of the rails on the left and right sides are exactly opposite, and the wear positions on the inner side appear alternately. Figure 3 and Figure 4 The scenes of alternating side grinding of the rails of the first example track segment and the second example track segment are respectively shown.

[0093] according to Figure 3 and Figure 4 The rail profile of the subway abnormal swaying area is obtained by using a mounted track inspection device, and the three-dimensional profiles of the rails in different longitudinal track sections are longitudinally connected to obtain the three-dimensional profile of the rails. Some results of the obtained three-dimensional profile of the rails can be found in Figure 5 shown.

[0094] In order to better ensure the smoothness of the alternating side grinding waveform, the alternating side grinding waveform corresponding to the left rail and the alternating side grinding waveform corresponding to the right rail can be obtained. In the process of obtaining the alternating side grinding waveform, in order to more intuitively display the changes in the rail profile during alternating side grinding, the three-dimensional profile of the rail is converted into an original height cloud map.

[0095] Figure 6 The original height cloud map corresponding to the left and right rails of 100 groups (within the mileage direction of 62.5m) is displayed. The horizontal direction in the original height cloud map represents the extension direction of the rail (i.e. the mileage direction), the vertical direction represents the horizontal direction perpendicular to the rail, and the color represents the vertical height of the rail. Figure 6 In order to better compare the left and right rail profiles, the original height cloud map corresponding to the left rail was shifted upward by 75mm, and the right rail was shifted downward by 75mm. Figure 6 , determine the side grinding area based on the original height cloud map, and then identify the highest peak and the lowest trough based on the side grinding area.

[0096] The original height cloud map is sliced ​​with the horizontal coordinate in the range of 0mm to 35mm and the interval of 5mm to form a curve of the vertical height changing with the longitudinal direction under different horizontal coordinate conditions, and recorded as the alternative waveform curve. The alternative waveform curve is as follows: Figure 7 However, since the waveforms of the alternative waveform curves are similar, but the wave depth amplitudes are significantly different, in order to more accurately describe the wave depth of the alternate side wear of the rail, it is necessary to find the horizontal coordinate at the maximum value of the profile height difference to determine the alternate side wear waveform.

[0097] By comparing the profiles of the highest crest and the lowest trough, it was determined that the longitudinal position of the highest crest was at 21.25m mileage, and the longitudinal position of the lowest trough was at 27.25m mileage. The variation characteristics of the vertical height with the lateral direction within the longitudinal position range were obtained, and the variation characteristics of the vertical height with the lateral direction within the longitudinal position range were analyzed to determine the horizontal coordinate of the maximum value of the profile height difference.

[0098] Figure 8 (a) shows the rail profile corresponding to the highest wave crest and the rail profile corresponding to the lowest wave trough. Figure 8 (b) shows the result of the profile height difference. Figure 8 As shown in (b), the height difference of the profile increases gradually with the increase of the horizontal coordinate y. When y = 33.1 mm, the height difference of the profile is the largest, reaching 5.1 mm. As the horizontal coordinate y continues to increase, the height difference of the profile drops sharply to zero. The waveform curve corresponding to y = 33.1 mm is used as the target waveform curve. The target waveform curve is then de-averaged to obtain the alternating side grinding waveform. The alternating side grinding waveform corresponding to the left rail and the alternating side grinding waveform corresponding to the right rail are shown in the figure. Fig. 9 shown.

[0099] like Fig.10 The figure shows the process of constructing the side grinding recognition model using the Swin Transformer algorithm. First, a multi-head self-attention (W-MSA) calculation process based on the window mechanism is implemented, and then a moving window multi-head self-attention (SW-MSA) calculation step is performed. Specifically, the input image Z l-1 First, the layer is normalized by layer normalization window (LN), and then W-WSA is applied to capture the features within the local window. After that, the output is normalized again and passed to a multi-layer perceptron (MLP) to achieve nonlinear changes. In this way, the construction of the basic transformation block is completed. Next, another multi-head self-attention mechanism based on a moving window is performed on the output to promote feature aggregation across windows. Then, the side grinding recognition model is constructed.

[0100] Based on the above method for detecting alternate side wear of rails, the present invention further provides a system for detecting alternate side wear of rails. The system for detecting alternate side wear of rails is used to implement the above method for detecting alternate side wear of rails.

[0101] Fig.11 Schematic diagram of the structure of the system for detecting alternate side wear of rails according to an embodiment of the present application. Fig.11 As shown, the system for detecting alternate rail side wear according to the embodiment of the present invention includes: a characteristic waveform generation module 1101 , a recognition model construction module 1102 and a real-time recognition module 1103 .

[0102] Specifically, the characteristic waveform generation module 1101 is implemented according to the method described in the above step S110, and is configured to obtain the three-dimensional profile of the rail profile in the abnormal swaying area of ​​the subway, and generate an alternating side grinding waveform at the side grinding position based on the three-dimensional profile of the rail; the recognition model construction module 1102 is implemented according to the method described in the above step S120, and is configured to construct a side grinding recognition model based on the alternating side grinding waveform; the real-time recognition module 1103 is implemented according to the method described in the above step S130, and is configured to collect the three-dimensional profile at different positions of the rail to be tested and generate corresponding alternating side grinding waveforms, and use the side grinding recognition model to determine the occurrence location of the alternating side grinding phenomenon of the rail.

[0103] The present invention proposes a method and system for detecting alternate side wear of rails. The method and system obtain the rail profile of the abnormal swaying area of ​​the subway, and use the rail profile to obtain a three-dimensional profile, which can more comprehensively capture the subtle changes on the rail surface and provide an accurate basis for the analysis of alternate side wear of rails; and after obtaining the three-dimensional profile, the alternating side wear waveform is obtained, which can provide an intuitive basis for analyzing the degree and distribution of alternate side wear of rails; finally, the side wear recognition model established can automatically and intelligently determine the occurrence location of the alternate side wear of rails. The present invention realizes the rapid and efficient identification and evaluation of alternate side wear of rails through the application of the side wear recognition model, which is of great significance for quantitatively evaluating the degree of alternate side wear of rails. Furthermore, it is expected to provide important technical support for subway operation maintenance and safety management.

[0104] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any person familiar with the technology within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0105] In the description of the present invention, unless otherwise specified, "plurality" means two or more than two; the orientations or positional relationships indicated by the terms "upper", "lower", "left", "right", "inner", "outer", "front end", "rear end", "head", "tail", etc. are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0106] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0107] It should be understood that the embodiments disclosed in the present invention are not limited to the specific structures, processing steps or materials disclosed herein, but should be extended to equivalent substitutions of these features understood by ordinary technicians in the relevant field. It should also be understood that the terms used herein are only used for the purpose of describing specific embodiments and are not meant to be limiting.

[0108] The "one embodiment" or "embodiment" mentioned in the specification means that the specific features, structures or characteristics described in conjunction with the embodiment are included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment" or "embodiment" appearing in various places throughout the specification do not necessarily refer to the same embodiment.

[0109] Although the embodiments disclosed in the present invention are as above, the contents described are only embodiments adopted for facilitating the understanding of the present invention and are not intended to limit the present invention. Any technician in the technical field to which the present invention belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in the present invention, but the patent protection scope of the present invention shall still be subject to the scope defined in the attached claims.

Claims

1. A method for detecting alternate side wear of rails, characterized in that: include: Obtain a three-dimensional profile of the rail profile in the area of ​​abnormal subway swaying, and generate an alternating side grinding waveform at the side grinding position according to the three-dimensional profile of the rail; constructing a side grinding recognition model according to the alternating side grinding waveform; The three-dimensional profiles at different positions of the rail to be tested are collected and corresponding alternating side wear waveforms are generated. The position where the alternating side wear phenomenon of the rail occurs is determined using the side wear recognition model.

2. The method according to claim 1, characterized in that The three-dimensional rail profile is obtained by the following steps: Using preset distance intervals, collect the three-dimensional profiles of rails in different longitudinal track sections in the subway abnormal swaying area; The three-dimensional profiles of the rails of different longitudinal track sections are longitudinally connected in series to obtain the three-dimensional profile of the rails, wherein the transverse direction of the three-dimensional profile of the rails represents the horizontal direction perpendicular to the rails, the longitudinal direction represents the extension direction of the rails, and the vertical direction represents the plumb line direction of the rails.

3. The method according to claim 2, characterized in that The step of generating an alternating side grinding waveform at a side grinding position according to a three-dimensional profile of the rail comprises: Converting the three-dimensional profile of the rail into an original height cloud map; Determine the abscissa at the maximum value of the height difference of the profile by identifying the highest peak and the lowest trough in the original height cloud map; Intercepting a height cloud map within a preset range and slicing the intercepted height cloud map to form a curve of vertical height changing with longitudinal direction under different horizontal coordinate conditions, which is recorded as an alternative waveform curve; According to the horizontal coordinate at the maximum value of the profile height difference and a plurality of alternative waveform curves, an alternating side grinding waveform is obtained.

4. The method according to claim 3, characterized in that The step of intercepting the height cloud map within a preset range and aligning and slicing it to form a curve of vertical height changing with the longitudinal direction under different horizontal coordinate conditions, which is recorded as an alternative waveform curve, includes: Generate an allowable value of the abscissa based on the abscissa at the maximum value of the height difference of the profile; Intercepting a height cloud map within a range between the horizontal coordinate allowable value and the track inner horizontal coordinate in the original height cloud map; The intercepted height cloud image is sliced ​​along the longitudinal direction at preset intervals to obtain a plurality of candidate waveform curves.

5. The method according to claim 3 or 4, characterized in that: The step of determining the abscissa at the maximum value of the profile height difference by identifying the highest peak and the lowest trough in the original height cloud map comprises: Determine the longitudinal position range between the longitudinal position of the highest wave peak and the longitudinal position of the lowest wave trough according to the highest wave peak and the lowest wave trough in the original height cloud map; Analyze the variation characteristics of vertical height with the lateral direction within the longitudinal position range, and determine the horizontal coordinate at the maximum height difference of the profile.

6. The method according to any one of claims 3 to 5, characterized in that: The step of obtaining an alternating side grinding waveform according to the horizontal coordinate at the maximum value of the profile height difference and a plurality of alternative waveform curves comprises: Based on the multiple candidate waveform curves, extract a curve of vertical height variation with longitudinal direction under the condition of the horizontal coordinate of the maximum value of the profile height difference, and record it as a target waveform curve; The target waveform curve is de-averaged to obtain the alternating side grinding waveform.

7. The method according to any one of claims 1 to 6, characterized in that: The alternating side grinding waveform includes: an alternating side grinding waveform corresponding to the left rail and an alternating side grinding waveform corresponding to the right rail.

8. The method according to claim 7, characterized in that According to the alternating side grinding waveform, the side grinding recognition model is constructed, which includes: According to the alternating side grinding waveform corresponding to the left rail and the alternating side grinding waveform corresponding to the right rail, in combination with an image processing method, an image to be input is obtained; According to the image to be input, combining the hierarchical feature processing method, obtaining input images corresponding to different levels; Using a preset shift window, the input image is shifted, an image corresponding to each shift window is intercepted, and feature analysis is performed on the intercepted image to obtain feature information corresponding to different levels; The side wear recognition model is constructed by using the feature information corresponding to all levels and the position of the rails as training data.

9. The method according to claim 8, characterized in that The levels include: low level, middle level and high level; Among them, the feature information corresponding to the low-level layer includes but is not limited to: edge features; The feature information corresponding to the intermediate level includes but is not limited to: shape features and structural features; The feature information corresponding to the high-level layer includes, but is not limited to: alternating side wear prediction result features.

10. A system for detecting alternate side wear of rails, characterized in that: include: A characteristic waveform generation module is configured to obtain a three-dimensional profile of a rail profile in an abnormally swaying area of ​​a subway and generate an alternating side grinding waveform at a side grinding position according to the three-dimensional profile of the rail; a recognition model building module configured to build a side grinding recognition model according to the alternating side grinding waveform; The real-time recognition module is configured to collect the three-dimensional profiles at different positions of the rail to be tested and generate corresponding alternating side wear waveforms, and determine the occurrence position of the alternating side wear phenomenon of the rail by using the side wear recognition model.

Images