A method for detecting absolute and relative vertical deformation of ballastless track structure

Abstract

The present invention discloses a method for detecting the vertical absolute and relative vertical deformation of a ballastless track structure. By only taking a side picture of the ballastless track structure that meets the requirements according to the algorithm requirements, the absolute displacement of the ballastless track structure and the relative displacement between each layer can be obtained conveniently and quickly. The algorithm has high calculation efficiency and accurate calculation results. It can adapt to different conditions such as roadbeds, bridges, and tunnels, and realizes non-contact non-destructive testing. The detection algorithm can help railway construction and operation and maintenance units to grasp the vertical deformation of multiple structural layers of the ballastless track in the measured section, and assist them in formulating targeted remediation plans. In addition, the detection results also provide data support for the study of deformation transfer laws of ballastless track systems.

Description

Technical Field

[0001] The invention relates to the field of ballastless track detection, and in particular to a method for detecting vertical absolute and relative vertical deformation of a ballastless track structure. Background Art

[0002] When high-speed railways pass through complex geological areas such as soft rocks, mining areas, karst caves, and frozen soil, the foundation under the ballastless track structure is prone to deformation. As the foundation deforms, the ballastless track structure system is prone to overall deformation following the foundation and relative deformation between the track structure layers. This deformation affects the long-term service health of the ballastless track structure and has an adverse effect on driving comfort and safety. In order to reduce the various adverse effects of the ballastless track structure caused by the deformation of the lower foundation, study the transmission law of the foundation deformation to the various layers of the ballastless track, and the transmission law of the displacement between the various layers of the ballastless track, it is urgent to accurately grasp the law of vertical displacement of the track with the deformation of the lower foundation, including the absolute displacement of the track structure relative to the CPIII control network leveling point, and the relative displacement between the track structure layers. However, the detection and calculation methods of the above related values ​​are not perfect at present.

[0003] Existing deformation detection of ballastless track structures is often only carried out during the operation and maintenance period, and the detection and subsequent calculation methods focus on the relative deformation of the structural layer, such as the gap detection scheme of air-coupled ultrasonic transmitting transducers, the track plate geometry detection scheme of CPIII control network combined with laser tracker, the gap detection trolley combined with machine vision algorithm, and the track plate three-dimensional position detection device combined with binocular ranging algorithm. None of the above equipment or algorithm methods can obtain the vertical absolute displacement of the ballastless track with the deformation of the foundation, and there are problems of low efficiency and complicated steps. At present, the absolute vertical deformation detection of ballastless track structure is still mainly obtained by repeatedly setting up stations with total stations combined with CPIII leveling control network. The problems of extremely low detection efficiency, long re-measurement cycle, and sparse measurement points along the line are particularly prominent. Summary of the invention

[0004] The technical problem to be solved by the present invention is the vertical absolute and relative displacement detection of ballastless track. To solve the above problem, the present invention adopts the following technical solutions:

[0005] A method for detecting absolute and relative vertical deformation of a ballastless track structure comprises the following steps:

[0006] Step S1: Positioning the section to be measured based on laser ranging data:

[0007] Step S2: Image preprocessing: performing image preprocessing on the image taken in the above step S1;

[0008] Perform image enhancement, image graying, and Gaussian filtering on the image taken in step S1 to improve image quality while reducing data volume, thereby reducing the amount of calculation;

[0009] Step S3: Image binarization based on the optimal threshold:

[0010] Perform threshold segmentation on the image obtained in step S2 above to obtain the optimal threshold, and then bring it into the binarization algorithm to obtain the optimal binarized image;

[0011] The maximum inter-class variance algorithm is used to perform threshold segmentation on the image obtained in the above step S2, and the optimal threshold is calculated; then the threshold is brought into the binarization algorithm, and the pixel points of the grayscale image are traversed one by one from left to right and from top to bottom. The pixel values ​​greater than the threshold are changed to 255, and the pixel values ​​less than the threshold are changed to 0, thereby obtaining the optimal binary image;

[0012] Step S4: image edge detection;

[0013] Perform edge detection on the image obtained in step S3 to obtain the edges of each layer structure on the side of the ballastless track structure and the edges of the horizontal laser line in the image;

[0014] Using the Canny edge detection algorithm, edge detection is performed on the image obtained in step S3 to obtain the edges of each layer structure on the side of the ballastless track structure and the edges of the horizontal laser line in the image;

[0015] Step S5: extracting main features of image edge information;

[0016] Performing main feature extraction on the image edge information extracted in step S4 above to remove useless edge information;

[0017] Step S6: Accurately locate the key edge line;

[0018] Perform edge line pixel point distribution feature analysis on the image obtained in step S5 to determine the position of the ballastless track structure edge and the horizontal laser line on the image;

[0019] Step S7: distinguishing the edge of the track structure and the edge of the laser line;

[0020] Step S8: Calculate the absolute and relative vertical displacements of the ballastless track structure between the two consecutive inspection dates.

[0021] Preferably, in step S3, the calculation steps of the maximum inter-class variance algorithm are as follows:

[0022] S3.1, assuming the threshold is K (0-255 is traversed in sequence, i.e. 0 is taken for the first calculation), and then the pixel values ​​of the pixels on the grayscale image are divided into two parts according to the K value:

[0023]

[0024] S3.2, calculate the pixel values ​​to become c 1 and c 2 The ratio of the number of pixels to the total number of pixels 1 and p 2 :

[0025]

[0026] p 2 =1-p 1 ;

[0027] S3.3, calculate the mean pixel value M of the original image:

[0028]

[0029]

[0030] S3.4, find the between-class variance σ 2 :

[0031] σ 2 =p 1 *(M c1 -M) 2 +p 2 *(M c2 -M) 2 =p 1 *p 2 *(M c1 -M c2 ) 2

[0032] Traverse the K value from 0 to 255, and find the value that makes σ 2 The maximum K value is denoted as K max , which is the target threshold.

[0033] Preferably, in step S3, the binarization algorithm is:

[0034] One by one through the image pixel values ​​M 0 , if M 0 >K max , M 0 becomes 255; if M 0 <K max , M 0 becomes 0.

[0035] Preferably, in step S4, the Canny edge detection algorithm calculation step includes:

[0036] S4.1, Gaussian filter

[0037] Assuming that the size of the Gaussian filter kernel is x×y and the standard deviation is σ, the Gaussian filter kernel matrix G is:

[0038]

[0039] In this example, the Gaussian filter kernel size is set to 3×3 and the standard deviation is 1. Substituting the above formula into the calculation, the Gaussian filter kernel is:

[0040]

[0041] Using the Gaussian filter kernel, convolution operations are performed on the binary image from left to right and from top to bottom;

[0042] S4.2, use the Sobel operator to calculate pixel gradient

[0043] The Sobel operator includes S x and S y , S x Used to calculate the pixel gradient matrix g in the x direction of the image x , S y Used to calculate the pixel gradient matrix g in the y direction of the image y ; The calculation method is:

[0044]

[0045]

[0046] Where I is the image pixel matrix, * represents the cross-correlation operation, the origin of the coordinate system is the upper left corner of the image, the positive direction of x is from left to right, and the positive direction of y is from top to bottom; then

[0047]

[0048] The gradient strength matrix g can be calculated xy ;

[0049] S4.3, non-maximum pixel gradient suppression

[0050] The pixel adjacency is divided into 4 regions, each of which contains two parts, the upper and lower parts. If the gradient intensity in the x direction of the central pixel is g x (i,j), the gradient strength in the y direction is g y (i,j), the gradient strength is g xy (i,j), then according to g x (i,j) and g yThe positive and negative and size of (i, j) can determine the area to which its gradient direction belongs, and then according to its pixel, according to its pixel gradient direction and the pixel gradient of the adjacent points, linear interpolation is performed to obtain the two gradient strengths g involved in the comparison of the positive and negative gradient directions. p1 (i,j) and g p2 (i,j); the formula is as follows:

[0051] g p1 (i,j)=(1-t)·g xy (i,j+1)+t·g xy (i-1,j+1)

[0052] g p2 (i,j)=(1-t)·g xy (i,j-1)+t·g xy (i+1,j-1)

[0053] If g xy (i,j)>g p1 (i,j) and g xy (i,j)>g p2 (i,j), the pixel is judged as an edge point and retained, otherwise it is a non-edge point and suppressed; the calculation methods for the other three areas are the same;

[0054] S4.4, dual threshold detection

[0055] Artificially set the pixel gradient value high threshold g High and g Low , judge all edge pixels one by one; if the gradient value of an edge pixel is greater than the high threshold, it is defined as a strong edge pixel; if it is less than the high threshold but greater than the low threshold, it is marked as a weak edge pixel; if the gradient value of an edge pixel is less than the low threshold, the pixel is suppressed;

[0056] S4.5, Suppression of isolated low threshold points

[0057] To obtain more accurate results, weak edge pixel values ​​caused by noise or color change should be suppressed. The specific method is to check the eight neighboring pixels of the weak edge pixel one by one. If one of them is a strong edge pixel, the weak edge pixel will be retained, otherwise it will be suppressed.

[0058] Preferably, in step S5,

[0059] The main features of the extracted image edge information in step S4 are extracted to automatically remove cluttered and useless edge information. The specific method is: through a self-compiled adaptive Canny edge detection high and low threshold automatic calculation method, the ideal edge detection high threshold and low threshold are obtained, and the two thresholds are brought into the original Canny edge detection algorithm to achieve the purpose of removing cluttered edges.

[0060] Preferably, in step S6, the image obtained in step S5 is further processed to accurately locate multiple structural edges on the side of the ballastless track structure in the image and horizontal laser lines with absolutely constant elevations, specifically, the pixel values ​​of the image are superimposed in units of lines, a threshold for determining whether the pixel superposition value is a structural edge or a line where the laser line is located is set, and the line numbers corresponding to the superposition values ​​greater than the threshold are screened out, and these line numbers are the lines where the ballastless track structure edges and horizontal laser lines are located on the image. Preferably, in step S7, specifically:

[0061] According to the characteristics of the pixel values, the position of the horizontal laser line on the original three-channel image taken by the camera is extracted, and combined with the result of step S7, the position of the edge line of the track structure and the position of the laser line are distinguished;

[0062] The rows where the horizontal laser lines are located on the original three-channel image are extracted according to the RGB value features, and the rows where the edge lines of the track structure are located and the rows where the laser lines are located are distinguished.

[0063] Preferably, in the step S8, specifically, the following steps are performed: analyzing the position difference between the structural edge line and the laser line of the same test section taken twice before and after, and the position difference between two different structural edge lines in the picture, and then using the correspondence between the track structure size in the image and the actual track structure size to obtain the vertical absolute displacement of the ballastless track structure twice before and after and the relative displacement between the structural layers.

[0064] Preferably, in step S1, specifically:

[0065] Using the laser equipment of the leveling point of the CPIII control network, project a laser horizontal line with a known elevation onto the side of the ballastless track structure, and use the ranging function of the mobile detection equipment to obtain real-time ranging data from the bottom of the mobile detection equipment to the surface entity of the ballastless track structure; determine the position of the mobile detection equipment on the track in real time during its movement;

[0066] Mobile detection equipment refers to the mobile detection frame, laser equipment refers to the self-leveling laser level surface emission module, and the laser equipment of the leveling point of the CPIII control network refers to the self-leveling laser level surface emission module rigidly connected to the CPIII elevation point embedded parts through rigid connectors;

[0067] When the distance measurement data shows the distance data feature from the bottom of the mobile inspection equipment to the surface entity of the ballastless track structure corresponding to the location of the section to be measured, it is considered that the equipment has reached the section to be measured. Then the camera on the mobile inspection equipment collects the side image of the multi-layer ballastless track structure to obtain a picture containing the edge features of the ballastless track side structure and a horizontal laser line with a known elevation.

[0068] Beneficial effects:

[0069] The present invention discloses a method for detecting the vertical absolute and relative vertical deformation of a ballastless track structure. By only taking a side picture of the ballastless track structure that meets the requirements according to the algorithm requirements, the absolute displacement of the ballastless track structure and the relative displacement between each layer can be obtained conveniently and quickly. The algorithm has high calculation efficiency and accurate calculation results. It can adapt to different conditions such as roadbeds, bridges, and tunnels, and realizes non-contact non-destructive testing. The detection algorithm can help railway construction and operation and maintenance units to grasp the vertical deformation of multiple structural layers of the ballastless track in the measured section, and assist them in formulating targeted remediation plans. In addition, the detection results also provide data support for the study of deformation transfer laws of ballastless track systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a flow chart of a method for detecting vertical absolute and relative vertical deformation of a ballastless track structure according to the present invention;

[0071] Figure 2 yes Figure 1 A schematic diagram of the overall application of the detection device structure used in step S1 in different basic sections;

[0072] Figure 3 yes Figure 2 Side view of the self-leveling laser leveling module installed on the bridge section;

[0073] Figure 4 yes Figure 2 Side view of the self-leveling laser leveling module installed on the roadbed section;

[0074] Figure 5 yes Figure 2 The front view of the self-leveling laser leveling surface transmitting module installed in the tunnel section;

[0075] Figure 6 yes Figure 2 Top view of the central universal retractable rigid cantilever structure;

[0076] Figure 7 yes Figure 2 Side view of the central universal retractable rigid cantilever structure;

[0077] Figure 8 It is the linear interpolation map of pixel gradient direction;

[0078] Fig. 9 This is a comparison chart of Canny edge detection effects;

[0079] Fig.10 It is the edge main feature extraction result image;

[0080] Fig.11 is a schematic diagram of displacement calculation results;

[0081] Explanation of the accompanying figures: 1. Mobile detection frame; 2. Distance measuring sensor; 3. Universally retractable rigid cantilever structure 4. Image acquisition module; 5. Data integration management system; 6. Drive module; 7. Rigid connector; 8. Self-leveling laser horizontal plane emission module; 9. Rail; 10. Track plate; 11. Self-compacting concrete; 12. Base plate; 13. Subgrade surface; 14. CPIII elevation embedded parts on the bridge crash barrier; 15. CPIII elevation embedded parts on the roadbed contact network pole; 16. CPIII elevation embedded parts on the tunnel lining. DETAILED DESCRIPTION

[0082] In order to more clearly illustrate the present invention, the present invention is further described below in conjunction with preferred embodiments and accompanying drawings. Similar components in the accompanying drawings are represented by the same reference numerals. It should be understood by those skilled in the art that the content specifically described below is illustrative rather than restrictive, and should not be used to limit the scope of protection of the present invention.

[0083] Figure 1 FIG. 1 is a flow chart of a method for detecting absolute and relative vertical deformation of a ballastless track structure according to the present invention. Figure 1 As shown, a method for detecting absolute and relative vertical deformation of a ballastless track structure of the present invention is described as follows:

[0084] Step S1: Positioning the section to be measured based on laser ranging data:

[0085] Using the laser equipment of the leveling point of the CPIII control network, project a laser horizontal line with a known elevation onto the side of the ballastless track structure, and use the ranging function of the mobile detection equipment to obtain real-time ranging data from the bottom of the mobile detection equipment to the surface entity of the ballastless track structure; determine the position of the mobile detection equipment on the track in real time during its movement;

[0086] The mobile detection device refers to the mobile detection frame 1 in the embodiment of the present invention (the specific structure is described in the subsequent description), the laser device refers to the self-leveling laser horizontal plane transmitting module 8, and the laser device of the leveling point of the CPIII control network refers to the self-leveling laser horizontal plane transmitting module 8 described in the subsequent embodiments, which is rigidly connected to the CPIII elevation point embedded part through the rigid connector 7;

[0087] When the distance data from the bottom of the mobile inspection device to the surface entity of the ballastless track structure corresponding to the position of the section to be measured appears in the distance measurement data, it is considered that the device has reached the section to be measured, and then the camera on the mobile inspection device collects the side image of the multi-layer ballastless track structure to obtain a picture containing the edge features of the ballastless track side structure and the horizontal laser line with a known elevation;

[0088] Step S2: Image preprocessing:

[0089] Performing image preprocessing on the image taken in step S1 above;

[0090] Perform image enhancement, image graying, and Gaussian filtering on the image taken in step S1 to improve image quality while reducing data volume, thereby reducing the amount of calculation;

[0091] Step S3: Image binarization based on the optimal threshold:

[0092] Perform threshold segmentation on the image obtained in step S2 above to obtain the optimal threshold, and then bring it into the binarization algorithm to obtain the optimal binarized image;

[0093] Use the maximum inter-class variance algorithm to perform threshold segmentation on the image obtained in step S2 above and calculate the optimal threshold. Then bring the threshold into the binarization algorithm, traverse the pixels of the grayscale image from left to right and from top to bottom one by one, and change the pixel values ​​greater than the threshold to 255, and the pixel values ​​less than the threshold to 0, so as to obtain the optimal binary image;

[0094] The calculation steps of the maximum between-class variance algorithm are as follows:

[0095] S3.1, assuming the threshold is K (0-255 is traversed in sequence, i.e. 0 is taken for the first calculation), and then the pixel values ​​of the pixels on the grayscale image are divided into two parts according to the K value:

[0096]

[0097] S3.2, calculate the pixel values ​​to become c 1 and c 2 The ratio of the number of pixels to the total number of pixels 1 and p 2 :

[0098]

[0099] p 2 =1-p 1

[0100] S3.3, calculate the mean pixel value M of the original image:

[0101]

[0102]

[0103] S3.4, find the between-class variance σ 2 :

[0104] σ 2 =p 1 *(M c1 -M) 2 +p 2 *(M c2 -M) 2 =p 1 *p 2 *(M c1 -M c2 ) 2

[0105] Traverse the K value from 0 to 255, and find the value that makes σ 2 The maximum K value is denoted as K max , which is the target threshold.

[0106] The calculation idea of ​​the binarization algorithm is as follows:

[0107] One by one through the image pixel values ​​M 0 , if M 0 >K max , M 0 becomes 255; if M 0 <K max , M 0 becomes 0.

[0108] Step S4: image edge detection;

[0109] Perform edge detection on the image obtained in the above step S3 to obtain the edges of each layer structure on the side of the ballastless track structure and the edges of the horizontal laser line in the image;

[0110] Using the Canny edge detection algorithm, edge detection is performed on the image obtained in step S3 to obtain the edges of each layer structure on the side of the ballastless track structure and the edges of the horizontal laser line in the image;

[0111] The calculation steps of the Canny edge detection algorithm are as follows:

[0112] S4.1, Gaussian filter

[0113] Assuming that the size of the Gaussian filter kernel is x×y and the standard deviation is σ, the Gaussian filter kernel matrix G is:

[0114]

[0115] In this example, the Gaussian filter kernel size is set to 3×3 and the standard deviation is 1. Substituting the above formula into the calculation, the Gaussian filter kernel is:

[0116]

[0117] Using the Gaussian filter kernel, convolution operations are performed on the binary image from left to right and from top to bottom.

[0118] S4.2, use the Sobel operator to calculate pixel gradient

[0119] The Sobel operator includes S x and S y , S x Used to calculate the pixel gradient matrix g in the x direction of the image x , S y Used to calculate the pixel gradient matrix g in the y direction of the image y The calculation method is:

[0120]

[0121]

[0122] Where I is the image pixel matrix, * represents the cross-correlation operation, the origin of the coordinate system is the upper left corner of the image, the positive x direction is from left to right, and the positive y direction is from top to bottom.

[0123]

[0124] The gradient strength matrix g can be calculated xy .

[0125] S4.3, non-maximum pixel gradient suppression

[0126] like Figure 8 As shown in the figure, the pixel adjacency is divided into 4 regions, each of which contains two parts, the upper and lower parts. If the gradient intensity in the x direction of the central pixel is g x (i,j), the gradient strength in the y direction is g y (i,j), the gradient strength is g xy (i,j), then according to g x (i,j) and g y The positive and negative and size of (i, j) can determine the area to which its gradient direction belongs, and then according to its pixel, according to its pixel gradient direction and the pixel gradient of the adjacent points, linear interpolation is performed to obtain the two gradient strengths g involved in the comparison of the positive and negative gradient directions. p1 (i,j) and g p2 (i,j).

[0127] The formula is as follows:

[0128] gp1 (i,j)=(1-t)·g xy (i,j+1)+t·g xy (i-1,j+1)

[0129] g p2 (i,j)=(1-t)·g xy (i,j-1)+t·g xy (i+1,j-1)

[0130] If g xy (i,j)>g p1 (i,j) and g xy (i,j)>g p2 (i,j), then the pixel is judged as an edge point and retained, otherwise it is a non-edge point and suppressed. The calculation methods for the other three regions are similar.

[0131] S4.4, dual threshold detection

[0132] Artificially set the pixel gradient value high threshold g High and g Low , all edge pixels are judged one by one. If the gradient value of an edge pixel is greater than the high threshold, it is defined as a strong edge pixel; if it is less than the high threshold but greater than the low threshold, it is marked as a weak edge pixel; if the gradient value of an edge pixel is less than the low threshold, the pixel is suppressed.

[0133] S4.5, Suppression of isolated low threshold points

[0134] To obtain more accurate results, weak edge pixel values ​​caused by noise or color change should be suppressed. The specific method is to check the eight neighboring pixels of the weak edge pixel one by one. If one of them is a strong edge pixel, the weak edge pixel will be retained, otherwise it will be suppressed.

[0135] The edge detection effect is as follows Fig. 9 As shown:

[0136] Step S5: extracting main features of image edge information;

[0137] Performing main feature extraction on the image edge information extracted in step S4 above to remove useless edge information;

[0138] The main features of the image edge information extracted in step S4 are extracted to automatically remove the cluttered and useless edge information. The specific method is: through the self-compiled adaptive Canny edge detection high and low threshold automatic calculation method, the ideal edge detection high threshold and low threshold are obtained, and the two thresholds are brought into the original Canny edge detection algorithm to achieve the purpose of removing cluttered edges;

[0139] The main feature extraction results of edge information are as follows Fig.10 As shown:

[0140] Step S6: Accurately locate the key edge line;

[0141] Perform edge line pixel point distribution feature analysis on the image obtained in step S5 to determine the position of the ballastless track structure edge and the horizontal laser line on the image;

[0142] The image obtained in the above step S5 is further processed to accurately locate multiple structural edges on the side of the ballastless track structure in the image and the horizontal laser line with absolutely constant elevation. The specific method is: superimpose the pixel values ​​of the image in units of rows, set a threshold for judging whether it is a structural edge or a pixel superposition value in the row where the laser line is located, and filter out the row numbers corresponding to the superposition values ​​greater than the threshold. These row numbers are the rows where the ballastless track structure edges and horizontal laser lines are located in the image.

[0143] Step S7: distinguishing the edge of the track structure and the edge of the laser line;

[0144] According to the characteristics of the pixel value, the position of the horizontal laser line on the original three-channel image taken by the camera is extracted, and combined with the result of step seven, the position of the edge line of the track structure and the position of the laser line are distinguished.

[0145] The row where the horizontal laser line (in this embodiment, the laser line is green) on the original three-channel image is extracted according to the RGB value feature, thereby distinguishing the row where the edge line of the track structure is located from the row where the laser line is located.

[0146] Step S8: Calculate the absolute and relative vertical displacements of the ballastless track structure between the two previous and next inspection dates;

[0147] The position differences between the structural edge lines and laser lines of the same test section taken twice in the images, as well as the position differences between two different structural edge lines in the images, are analyzed. Then, the correspondence between the track structure size in the image and the actual track structure size can be used to obtain the vertical absolute displacement of the ballastless track structure twice and the relative displacement between the structural layers.

[0148] Calculate the difference in the number of vertical pixels (i.e., the difference in row numbers) between the key lines at the same test section taken at two different times, recorded as n. Then, according to the following (Equation 1), the vertical absolute displacement and relative displacement of each layer of the ballastless track structure before and after the two times can be obtained, recorded as DIS.

[0149]

[0150] In formula (1):

[0151] DIS—vertical absolute and relative displacement of each layer of ballastless track structure;

[0152] H—total thickness of ballastless track structure;

[0153] α—the percentage of the ballastless track structure thickness in the image in the vertical direction (controllable by humans);

[0154] N—the number of vertical pixels of the image;

[0155] n—The difference in the row number of the key line at the same position taken at two different times, that is, the difference in the number of vertical pixels. When calculating the relative displacement between two structural layers of the ballastless track, the key line correspondingly selects the two ballastless track structural edges to be calculated; when calculating the absolute displacement of a structural layer of the ballastless track, the key line correspondingly selects the ballastless track structural edge to be calculated and the laser line.

[0156] The results of steps S6-S8 are as follows: Fig.11 As shown:

[0157] Steps S2-S8 are all completed by the data integration management system 5.

[0158] The device structure used in the detection method described in the above embodiment is described as follows:

[0159] Figure 2 It is a general schematic diagram of the application of the detection device of the present invention in different basic sections, including three situations: bridge section, roadbed section, and tunnel section. Figure 2 As shown, the ballastless track structure vertical absolute and relative vertical deformation detection device of this embodiment includes a mobile detection frame 1, a distance sensor 2, a universal retractable rigid cantilever structure 3, an image acquisition module 4, a data integration management system 5, a drive module 6, a rigid connector 7, and a self-leveling laser horizontal plane emission module 8. The chassis size of the mobile detection frame 1 is 0.9m×1.6m×0.03m in length×width×thickness. The mobile detection frame 1 is made of light alloy material and is coupled with the rail through multiple sets of nylon wheel pairs. The drive module 6 is driven by a hybrid drive mode of electric drive and hand push.

[0160] A laser distance sensor 2 is fixed under the chassis on one side of the mobile detection frame 1. The measuring point is placed on the surface of the track plate and the moving track passes above the rail support platform on the same side. High-frequency sampling can be achieved during the mobile monitoring process. The distance sensor 2 can be connected to the data integration management system 5 by wire or wireless mode.

[0161] Combination Figure 6 , Figure 7As shown, a universally retractable rigid cantilever structure 3 is fixed on each side of the chassis of the mobile detection frame 1, and the maximum extension length of the multi-section retractable cantilever rod is 2m; a small image acquisition module 4 integrated with a 16-megapixel camera and an adaptive fill light component is installed at a fixed interface at the end of the cantilever structure to obtain the side image of the track plate and the shooting time and position information; the universally retractable rigid cantilever structure 3 of the present invention is respectively arranged on both sides of the chassis of the mobile detection frame 1, and is connected to the chassis through a universal rotating structure, the cantilever rod of the universally retractable rigid cantilever structure 3 is a lightweight rigid structure, and the image acquisition module 4 is installed at the end of the retractable and fixed cantilever, the cantilever rod can be retracted, and there are multiple equipment fixing interfaces with universal rotation function at the end of the cantilever rod. The retraction and rotation of the universally retractable rigid cantilever structure 3 and the rotation behavior of the equipment structure can be switched and controlled manually or by the data integration management system 5;

[0162] Combination Figure 3 , Figure 4 , Figure 5 As shown, on the same side where the image acquisition module 4 is installed, a self-leveling laser horizontal plane transmitting module 8 is rigidly connected to the CPIII elevation point embedded part through a rigid connector 7, forming a horizontal laser line with a length of not less than 60m and a known height difference with the embedded part on the side of the ballastless track structure; the rigid connector 7 is composed of a threaded rod of a rigid material with known size and a lightweight connecting rod. One end of the bolt rod is smooth, and the thread and size of the other end are precisely machined according to the requirements of the CPIII prism rod to ensure that the bolt is screwed into the embedded part of the CPIII elevation control, and the error meets the requirements of track precision control measurement. The annular end of the connecting rod is nested in the smooth end of the threaded rod to keep the connecting rod freely hanging, and the cylindrical thread at the other end of the connecting rod is screwed into the threaded hole on the top of the self-leveling laser horizontal plane transmitting module 8 to fix the laser horizontal plane transmitting module. Rigid connectors of different structural forms and sizes are used for different high-speed railway foundation forms;

[0163] A data integration management system 5 is fixed on the mobile detection frame 1, and the data integration management system 5 is undertaken by a high-computing-power micro-board system or a remote data terminal. The high-computing-power micro-board system is fixed on the support of the chassis of the detection device frame and supports portable power supply; the data integration management system 5 includes a mechanical control module, which controls the spatial position and orientation of the interface fixing device by adjusting the cantilever rod length, the base and the rotation angle of the device fixing interface of the universal retractable rigid cantilever structure 3 in a wired or wireless manner. At the same time, the mechanical control module is connected to the electric drive device of the drive module 6 to adjust the rotation speed and direction of the detection frame wheelset;

[0164] The data integrated management system 5 of the present invention includes a measuring point positioning module, which stores and processes the ranging value obtained by the ranging sensor 2 in real time, stores the ranging data in real time, analyzes and processes the ranging data, and counts the time and number of periodic increases and decreases of the ranging value, thereby realizing detection mileage positioning in units of the longitudinal spacing of the rail support platform.

[0165] The data integration management system 5 includes an image recognition processing module, which receives the real-time video stream transmitted by the image acquisition module 4 in real time, automatically identifies and takes side high-definition pictures when the mobile detection frame 1 moves to the ends of the track plate and the base plate, and obtains the relative height difference between the side edge line of the track structure and the laser horizontal line through image time extraction, analysis and processing.

[0166] The data integration management system 5 includes a data fusion processing module, which performs time alignment and synchronous trigger timing on the image acquisition module 4 and the ranging sensor 2, monitors the travel speed of the detection frame to adjust the sampling frequency of the ranging sensor, the shooting frequency and pixels of the image acquisition module in real time.

[0167] The described embodiments are only a part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

Claims

1. A method for detecting absolute and relative vertical deformation of a ballastless track structure, characterized in that: The following steps are included: Step S1: Positioning the section to be measured based on laser ranging data: Step S2: Image preprocessing: performing image preprocessing on the image taken in the above step S1; Perform image enhancement, image graying, and Gaussian filtering on the image taken in step S1 to improve image quality while reducing data volume, thereby reducing the amount of calculation; Step S3: Image binarization based on the optimal threshold: Perform threshold segmentation on the image obtained in step S2 above to obtain the optimal threshold, and then bring it into the binarization algorithm to obtain the optimal binarized image; The maximum inter-class variance algorithm is used to perform threshold segmentation on the image obtained in the above step S2, and the optimal threshold is calculated; then the threshold is brought into the binarization algorithm, and the pixel points of the grayscale image are traversed one by one from left to right and from top to bottom. The pixel values ​​greater than the threshold are changed to 255, and the pixel values ​​less than the threshold are changed to 0, thereby obtaining the optimal binary image; Step S4: image edge detection; Perform edge detection on the image obtained in step S3 to obtain the edges of each layer structure on the side of the ballastless track structure and the edges of the horizontal laser line in the image; Using the Canny edge detection algorithm, edge detection is performed on the image obtained in step S3 to obtain the edges of each layer structure on the side of the ballastless track structure and the edges of the horizontal laser line in the image; Step S5: extracting main features of image edge information; Performing main feature extraction on the image edge information extracted in step S4 above to remove useless edge information; Step S6: Accurately locate the key edge line; Perform edge line pixel point distribution feature analysis on the image obtained in step S5 to determine the position of the ballastless track structure edge and the horizontal laser line on the image; Step S7: distinguishing the edge of the track structure and the edge of the laser line; Step S8: Calculate the absolute and relative vertical displacements of the ballastless track structure between the two consecutive inspection dates.

2. A method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: In step S3, the calculation steps of the maximum between-class variance algorithm are as follows: S3.1, assuming the threshold is K (0-255 is traversed in sequence, that is, O is taken for the first calculation), and then the pixel values ​​of the pixels on the grayscale image are divided into two parts according to the K value: S3.2, respectively calculate the proportions p1 and p2 of the number of pixels whose pixel values ​​become c1 and c2 to the total number of pixels: p2=1-p1; S3.3, calculate the mean pixel value M of the original image: S3.4, find the between-class variance σ 2 : σ 2 =p1*(M c1 -M) 2 +p2*(M c2 -M) 2 =p1*p2*(M c1 -M c2 ) 2 Traverse the K value from 0 to 255 to find the value that makes σ 2 The maximum K value is denoted as K max , which is the target threshold.

3. The method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: In the step S3, the binarization algorithm is: Go through the image pixel values ​​M0 one by one, if M0>K max , change M0 to 255; if M0<K max , change M0 to O.

4. The method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: In step S4, the Canny edge detection algorithm calculation step includes: S4.1, Gaussian filter Assuming that the size of the Gaussian filter kernel is x×y and the standard deviation is σ, the Gaussian filter kernel matrix G is: In this example, the Gaussian filter kernel size is set to 3×3 and the standard deviation is 1. Substituting the above formula into the calculation, the Gaussian filter kernel is: Using the Gaussian filter kernel, convolution operations are performed on the binary image from left to right and from top to bottom; S4.2, use the Sobel operator to calculate pixel gradient The Sobel operator includes S x and S y , S x Used to calculate the pixel gradient matrix g in the x direction of the image x , S y Used to calculate the pixel gradient matrix g in the y direction of the image y ; The calculation method is: Where I is the image pixel matrix, * represents the cross-correlation operation, the origin of the coordinate system is the upper left corner of the image, the positive direction of x is from left to right, and the positive direction of y is from top to bottom; then The gradient strength matrix g can be calculated xv ; S4.3, non-maximum pixel gradient suppression The pixel adjacency is divided into 4 regions, each of which contains two parts, the upper and lower parts. If the gradient intensity in the x direction of the central pixel is g x (i, j), the gradient strength in the y direction is g y (i, j), the gradient strength is g xy (i, j), then according to g x (i, j) and g y The positive and negative and size of (i, j) can determine the area to which its gradient direction belongs, and then according to its pixel, the linear interpolation of its pixel gradient direction and the pixel gradient of the adjacent points can obtain the two gradient strengths g involved in the comparison of the positive and negative gradient directions. p1 (i, j) and g p2 (i, j); the formula is as follows: g p1 (i,j)=(1-t)·g xy (i,j+1)+t·g xy (i-1,j+1) g p2 (i,j)=(1-t)·g xy (i,j-1)+t·g xy (i+1,j-1) If g xy (i, j)>g p1 (i, j) and g xy (i, j)>g p2 (i, j), the pixel is judged as an edge point and retained, otherwise it is a non-edge point and suppressed; the calculation methods for the other three areas are the same; S4.4, dual threshold detection Artificially set the pixel gradient value high threshold g High and g Low , judge all edge pixels one by one; if the gradient value of an edge pixel is greater than the high threshold, it is defined as a strong edge pixel; if it is less than the high threshold but greater than the low threshold, it is marked as a weak edge pixel; if the gradient value of an edge pixel is less than the low threshold, the pixel is suppressed; S4.5, Suppression of isolated low threshold points To obtain more accurate results, weak edge pixel values ​​caused by noise or color changes should be suppressed; the specific method is to check the eight neighboring pixels of the weak edge pixel one by one. As long as one of them is a strong edge pixel, the weak edge pixel will be retained, otherwise it will be suppressed.

5. The method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: In the step S5, The main features of the extracted image edge information in step S4 are extracted to automatically remove cluttered and useless edge information. The specific method is: through a self-compiled adaptive Canny edge detection high and low threshold automatic calculation method, the ideal edge detection high threshold and low threshold are obtained, and the two thresholds are brought into the original Canny edge detection algorithm to achieve the purpose of removing cluttered edges.

6. A method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: In the step S6, the image obtained in the step S5 is further processed to accurately locate the multiple structural edges on the side of the ballastless track structure in the image and the horizontal laser line with absolutely constant elevation. Specifically, the pixel values ​​of the image are superimposed in units of rows, and a threshold for determining whether the pixel superposition value is a structural edge or a row where the laser line is located is set. The row numbers corresponding to the superposition values ​​greater than the threshold are screened out. These row numbers are the rows where the ballastless track structure edges and horizontal laser lines are located in the image.

7. A method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: The step S7 is specifically as follows: According to the characteristics of the pixel values, the position of the horizontal laser line on the original three-channel image taken by the camera is extracted, and combined with the result of step S7, the position of the edge line of the track structure and the position of the laser line are distinguished; The rows where the horizontal laser lines are located on the original three-channel image are extracted according to the RGB value features, and the rows where the edge lines of the track structure are located and the rows where the laser lines are located are distinguished.

8. The method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: In the step S8, specifically, the position difference between the structure edge line and the laser line of the same test section taken twice before and after, and the position difference between two different structure edge lines in the picture are analyzed, and then the vertical absolute displacement of the ballastless track structure twice before and after and the relative displacement between the structural layers are obtained respectively by using the correspondence between the track structure size in the image and the actual track structure size.

9. A method for detecting absolute and relative vertical deformation of a ballastless track structure according to claim 1, characterized in that: In the step S1, specifically: Using the laser equipment of the leveling point of the CPIII control network, project a laser horizontal line with a known elevation onto the side of the ballastless track structure, and use the distance measurement function of the mobile detection equipment to obtain real-time distance measurement data from the bottom of the mobile detection equipment to the surface entity of the ballastless track structure; Real-time determination of the position of the mobile inspection equipment on the track during its travel; Mobile detection equipment refers to the mobile detection frame, laser equipment refers to the self-leveling laser level surface emission module, and the laser equipment of the leveling point of the CPIII control network refers to the self-leveling laser level surface emission module rigidly connected to the CPIII elevation point embedded parts through rigid connectors; When the distance measurement data shows the distance data feature from the bottom of the mobile inspection equipment to the surface entity of the ballastless track structure corresponding to the location of the section to be measured, it is considered that the equipment has reached the section to be measured. Then the camera on the mobile inspection equipment collects the side image of the multi-layer ballastless track structure to obtain a picture containing the edge features of the ballastless track side structure and a horizontal laser line with a known elevation.

Images