Following type contact line geometric parameter measuring method and device

Abstract

The invention discloses a following type contact line geometric parameter measuring method and device, and belongs to the technical field of railway detection, and the method mainly comprises the steps: S1, collecting an image containing a contact line through a measuring head; s2, segmenting the form of a contact line in the image, and extracting a contact line region; s3, based on the contact line area, calculating a pixel center line of the contact line; s4, calculating a vertical physical distance between the pixel center line of the contact line and the preset reference pixel point; s5, calculating a target rotation angle of the measuring head according to the vertical physical distance; s6, driving the measuring head to rotate to the target rotation angle, and measuring and calculating contact line geometric parameters of the current measuring point; and S7, repeating the steps S1 to S6, and completing continuous measurement of the geometric parameters of the contact line of the to-be-measured area. According to the invention, automatic, continuous and high-precision tracking measurement of the moving contact line can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of railway detection technology, in particular to a following type contact line geometric parameter measurement method and device. BACKGROUND

[0002] The railway contact line is an important part of the electrified railway, responsible for providing power for the railway electric locomotive, and the state of the contact line directly affects the running quality of the train on the line, so it is the focus of the railway management department. The geometric parameters of the railway contact line mainly include the sag and the height of the contact line at the positioning point, and the pull-out value, wherein the height of the contact line is the height of the contact line, and the pull-out value is the horizontal distance of the contact line relative to the track center line, and the accuracy of the height and the pull-out value is directly related to the smooth current collection of the pantograph and the safety of train operation, therefore, periodic and efficient measurement of the geometric parameters of the contact line is an important work of the railway operation and maintenance department.

[0003] At present, the measurement methods of the contact line geometric parameters mainly include vehicle-mounted device measurement and portable device measurement, among which the portable device measurement has become the mainstream, which usually requires the measurement personnel to push the portable measuring instrument directly below the measurement point, manually operate for artificial point measurement, and then operate the measurement button of the instrument to obtain the corresponding measurement value. This method has obvious shortcomings: 1. The artificial point measurement obtains discrete measurement point data of the contact line geometric parameters, and cannot obtain continuous measurement data, which cannot reflect the continuous change of the contact line parameters of the whole line, and may miss the detection of local defects; 2. The measurement efficiency is low, the labor intensity is large, and it is difficult to adapt to the rapid detection demand of long distance and large range.

[0004] In order to realize continuous measurement, the key is to make the light point of the measuring device, such as the laser range finder, continuously and accurately aim at the center of the contact line during movement. If the light point cannot aim at the center of the contact line, for example, only half of it is irradiated on the contact line, then the distance value fed back by the laser range finder is not the straight line distance from the true light center of the laser range finder to the center of the contact line, but the wrong distance to the edge of the contact line or other positions, which will directly lead to serious deviation in the calculation of the height and the pull-out value.

[0005] Further, since the contact line is arranged in zigzag shape on the line, and as the measurement trolley runs on the track, the measurement head of the laser range finder needs to have the rotation alignment capability to ensure that the laser range finder always aims at the center of the contact line. However, the existing scheme mainly adopts the method of manually adjusting the angle of the measurement head to align the position of the laser light point on the contact line. This manual alignment method has obvious problems: First, the alignment process is cumbersome and time-consuming, which severely restricts measurement efficiency and makes continuous measurement difficult to achieve. Second, during the implementation process, operators need to make frequent adjustments, making it impossible for them to focus on equipment implementation and environmental observation, which poses safety hazards. Finally, manual judgment is subject to subjective errors, making it difficult to guarantee alignment accuracy.

[0006] Therefore, there is an urgent need for a means to automatically, accurately, and in real time align the contact wire and perform continuous geometric parameter measurements. Summary of the Invention

[0007] The purpose of this invention is to overcome the problems existing in the prior art and to provide a method and apparatus for measuring the geometric parameters of a follow-up contact line.

[0008] The objective of this invention is achieved through the following technical solution: Firstly, a method for measuring the geometric parameters of a follow-up contact line is provided, including: Initial alignment measurement steps: Use the measuring head to obtain the initial geometric parameters of the contact line at the first measuring point in the area to be measured; Automatic follow-up continuous measurement steps: Perform the following process for subsequent measurement points: S1. Use a measuring head to acquire an image containing the contact wire; S2. Segment the shape of the contact line in the image and extract the contact line region; S3. Based on the contact line region, calculate the pixel center line of the contact line; S4. Calculate the vertical physical distance between the pixel center line of the contact line and the preset reference pixel; S5. Calculate the target rotation angle of the measuring head based on the vertical physical distance; S6. Drive the measuring head to rotate to the target rotation angle, measure and calculate the contact line geometric parameters of the current measuring point; S7. Repeat steps S1-S6 to complete the continuous measurement of the contact wire geometric parameters of the area to be measured.

[0009] In some embodiments, a deep learning segmentation network is used in step S2 to segment the shape of the contact line in the image.

[0010] In some embodiments, the deep learning segmentation network employs the Unet network, whose loss function is: Loss = α·BCE + β·DCE + γ·FL, where BCE represents the basic binary cross-entropy loss, DCE represents the overlap loss for the edge region, FL represents the loss for the width of the blurred edge region, and α, β, and γ are weighting coefficients.

[0011] In some embodiments, the pixel center line of the contact line is calculated using the pixel center method in step S3.

[0012] In some embodiments, calculating the pixel center line of the contact line using the pixel center method includes: Take multiple contact wire sections at the edge of the contact wire area; Determine the left and right edge points of the contact wire cross section, and take the midpoint of the line connecting the left and right edge points as the center point of the contact wire cross section; The center points of multiple contact wire sections are connected to form the pixel center line.

[0013] In some embodiments, calculating the pixel center line of the contact line using the pixel center method further includes: The pixel centerline is fitted using a least-squares linear fitting algorithm: Remove the center points of the contact wire cross-sections at preset ratios before and after the image; The center points of the remaining contact wire sections are merged into a point set, and a straight line equation is used for fitting to obtain the fitted pixel center line.

[0014] In some embodiments, when fitting using a linear equation, the sum of squares of the vertical distance errors from all data points to the fitted line is minimized.

[0015] In some embodiments, the preset reference pixel is the center point of the image.

[0016] In some embodiments, the target rotation angle is calculated using the following formula: Δa = arctan(H / L) - arctan(H / (L+ΔL)), where H is the guide height value calculated from the previous measuring point, L is the pull-out value calculated from the previous measuring point, and ΔL is the vertical physical distance.

[0017] Secondly, a follow-type contact wire geometric parameter measuring device is provided, comprising: A traveling mechanism for moving on a track; A measuring frame, connected to the traveling mechanism; A rotatable measuring head is mounted on the measuring frame, and the rotatable measuring head integrates a camera and a laser rangefinder; The control unit is electrically connected to the camera, laser rangefinder, and drive mechanism; the control unit is configured to perform the method described in the first aspect.

[0018] It should be further noted that the technical features corresponding to the above-mentioned options and embodiments can be combined or substituted with each other to form new technical solutions without conflict.

[0019] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention can identify the position of the contact wire in real time and automatically adjust the angle of the measuring head so that the laser rangefinder spot is continuously aligned with the center of the contact wire. This allows for the automatic and continuous collection of guide height and pull-out value data during the device's advancement, overcoming the shortcomings of discrete data from traditional manual single-point measurements and providing a comprehensive reflection of the line status.

[0020] 2. This invention uses a fully automated tracking and alignment process to replace inefficient and cumbersome manual observation and adjustment. Operators only need to push the equipment without frequent instrument operation, which not only doubles the measurement speed and enables rapid long-distance detection, but also allows personnel to focus more on the pushing path and environmental safety, making the measurement of contact wire geometric parameters more efficient and safer.

[0021] 3. This invention employs an image segmentation method based on deep learning (such as UNet), capable of adapting to various complex lighting and environmental scenarios, including sunny days, cloudy days, nighttime, and tunnels, accurately segmenting contact line edges. Combined with an optimized centerline extraction and fitting algorithm, sub-pixel accuracy in position recognition is ensured. Automatic alignment avoids subjective human error, guaranteeing that the laser rangefinder always operates at its optimal state, thereby improving the accuracy of geometric parameter calculations from the source. Attached Figure Description

[0022] Figure 1 This is a flowchart of a method for measuring the geometric parameters of a follow-up contact wire according to the present invention; Figure 2 This is a schematic diagram of the structure of a follower-type contact wire geometric parameter measuring device according to the present invention; Figure 3 This is a schematic diagram illustrating the geometric parameter measurement principle of the present invention. Detailed Implementation

[0023] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] It should be noted that the defects in the solutions in the prior art are all the results of the inventors' practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the embodiments of this application in the following text should be the inventors' contributions to this application in the process of invention and creation, and should not be understood as technical content known to those skilled in the art.

[0025] In view of the technical problems pointed out in the background art, the present invention provides the following embodiments: like Figure 2 As shown, a follow-type contact wire geometric parameter measuring device (measuring trolley) is provided, comprising: The traveling mechanism (including 3 traveling wheels 1 in the figure) is used for moving on the track; Measuring frame 2 is connected to the walking mechanism; A rotatable measuring head 4 is mounted on the measuring frame, and the rotatable measuring head 4 integrates a camera and a laser rangefinder; for example... Figure 2 As shown, the rotatable measuring head 4 is mounted on one side (not in the middle) of the measuring frame 2 via an electric rotary drive device 5 (drive mechanism).

[0026] The control unit, with an embedded structure, is electrically connected to the camera, laser rangefinder, and drive mechanism; the control unit is configured to perform a follow-type contact line geometry parameter measurement method.

[0027] The rotating measuring head 4 is equipped with a supplementary light, a push rod 3, an encoder, and a track gauge measuring unit.

[0028] In another exemplary embodiment, based on the above-described apparatus, a method for measuring the geometric parameters of a following contact line is provided, comprising: Initial alignment measurement steps: Place the measuring carriage on the track, manually rotate the measuring head to align the image center point with the contact line, click the measurement data button, and use the measuring head to acquire the initial geometric parameters of the contact line at the first measuring point in the area to be measured, completing the first measurement. The measurement principle is as follows: Figure 3 As shown, O represents the position of the contact line at a certain measurement moment. A signal beam is emitted from the measuring head 4 to measure the straight-line distance X between it and the contact line. The rotation angle α of the measuring head 4 is obtained by the encoder on the drive device. The distance between the measuring head 4 and the track centerline is obtained by the track gauge measuring unit of the trolley, and this value is recorded as L1. From this, we can obtain H=Xsin(a); L=Xcos(a) through trigonometric function relationships. H is the guide height value of the contact line at this measurement point, and L-L1 is the pull-out value of the contact line at this measurement point.

[0029] Then, click the "Follow Measurement" button to activate the automatic follow mode. Since the contact line is arranged in a zigzag pattern on the track, as the measuring trolley moves along the track, the trolley needs to continuously adjust the angle 'a' of the measuring head to ensure that the rangefinder of measuring head 4 is always aligned with the contact line, in order to obtain continuous X values, and thus continuous contact line guide height and pull-out parameters. When the measuring trolley measures the contact line geometric parameters at different positions, the rotation angle of measuring head 4 is different. Therefore, how to enable the laser rangefinder in measuring head 4 to automatically and accurately track the constantly changing position of the contact line, thereby obtaining the straight-line distance between measuring head 4 and the contact line, is the key technology of this invention. Specifically, the automatic follow continuous measurement steps are as follows: Figure 1 As shown, the following procedure is performed on subsequent measuring points: S1. Use measuring head 4 to acquire an image containing the contact wire; S2. Segment the shape of the contact line in the image and extract the contact line region; S3. Based on the contact line region, calculate the pixel center line of the contact line; S4. Calculate the vertical physical distance between the pixel center line of the contact line and the preset reference pixel; S5. Calculate the target rotation angle of the measuring head based on the vertical physical distance; S6. Drive the measuring head 4 to rotate to the target rotation angle, measure and calculate the contact line geometric parameters of the current measuring point; S7. Repeat steps S1-S6 to complete the continuous measurement of the contact wire geometric parameters of the area to be measured.

[0030] In step S1, images are continuously acquired using a camera in the measuring head. In step S2, the Unet deep learning network is used to segment the contact line morphology in the images, specifically including: 1. Dataset Construction: Utilizing the current camera's imaging state, acquire contact line images in various scenarios including sunny days, cloudy days, nighttime, mountainous areas, tunnels, straight lines, and curved lines. Perform pixel-level edge annotations on the contact lines in the images, ensuring each annotation edge is at least 2 pixels wide. The dataset should contain at least 2000 images. Then, perform image preprocessing, including image normalization, resizing, and noise reduction. For motion-blurred images, use bilateral filtering for smoothing; for uneven lighting, use CLAHE to enhance edge contrast. 2. Optimize the Unet model to improve edge localization accuracy. Specifically, this optimization involves designing a loss function using a combined loss method. The loss function is: Loss = α·BCE + β·DCE + γ·FL, where BCE represents the basic binary cross-entropy loss, used to balance positive and negative samples; DCE represents the overlap loss for edge regions, addressing the imbalanced sample problem; FL represents the loss for the width of blurred edge regions; and α, β, and γ are weighting coefficients, with α + β + γ = 1. These values ​​are fine-tuned based on validation results, with default values ​​of α = 0.2, β = 0.4, and γ = 0.4.

[0031] 3. Unet model training, monitoring, and parameter tuning; 4. Deployment and application of the Unet model.

[0032] Preferably, the step S3, which calculates the pixel center line of the contact line using the pixel center method, specifically includes: Take multiple contact wire sections at the edge of the contact wire area; Determine the left and right edge points of the contact wire cross section, and take the midpoint of the line connecting the left and right edge points as the center point of the contact wire cross section; Connecting the center points of multiple contact wire cross-sections forms the pixel centerline. For example, in the vertical direction of the image, for each row (or several rows) of pixels, find the left and right edge pixel coordinates of the contact line area, and take the midpoint as the center point of that row. Connect the center points of all rows to form the initial pixel centerline.

[0033] Then, to improve robustness, a least-squares linear fitting algorithm is used to fit the center line of the pixel: Remove the center points of the contact wire cross-sections at preset ratios before and after the image; The center points of the remaining contact wire sections are merged into a point set, and a straight line equation is used to fit the set to obtain the fitted pixel center line. For example, the first and last 10% of the points in the initial center point sequence are removed (these points may be located at the image edge and are more susceptible to interference). For the remaining center point set {( , We fit a straight line equation y=kx+b using the least squares method to minimize the sum of squared errors. This fitted line is the final smooth contact line centerline used for calculation.

[0034] In step S4, the vertical pixel distance between the fitted center line and the image center point (preset reference point) in the horizontal direction is calculated. Combining the known contact line height information and the camera calibration parameters, the pixel resolution of the contact line at the current distance is calculated. Multiplying the pixel distance by the pixel resolution yields the vertical physical distance ΔL. ΔL can be positive or negative depending on the configuration requirements. For example, if the configuration requires the contact line to be positive to the left of a specified point and negative to the right, the calculated value should conform to the configuration requirements, and vice versa.

[0035] In step S5, the current target rotation angle is calculated by combining the vertical physical distance, camera parameters, and geometric parameters measured at the previous measurement point: Δa = arctan(H / L) - arctan(H / (L+ΔL)), where H is the guide height value calculated from the previous measuring point, L is the pull-out value calculated from the previous measuring point, and ΔL is the vertical physical distance.

[0036] In step S6, the electric rotary drive device 5 is controlled to complete the rotation according to the target rotation angle requirement. After the measuring head 4 rotates to the correct position, the laser rangefinder immediately measures the new distance, the encoder reads the new angle, and calculates and stores the guide height and pull-out value of the current point accordingly.

[0037] The measuring carriage is pushed by push rod 3 and traveling wheels 1 to complete the measurement of all measuring points in the preset area. Clicking the end button saves the data and completes the measurement. In this way, the device can automatically maintain the tracking of the laser rangefinder spot to the center of the contact line during the pushing process, without manual intervention, and efficiently and continuously complete the acquisition of the geometric parameters of the entire contact line.

[0038] Furthermore, in another example, the initial alignment measurement step is only used for initial alignment and not as the measurement result for the first measuring point. After the initial alignment, the measurement of the first measuring point is initiated, and the calculation of the first measuring point is based on the data measured in the initial alignment.

[0039] The above detailed embodiments are a description of the present invention. It should not be considered that the specific embodiments of the present invention are limited to these descriptions. For those skilled in the art, several simple deductions and substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the protection scope of the present invention.

Claims

1. A method for measuring geometric parameters of a trolley line, characterized in that, The method comprises the following steps: An initial alignment measurement step: using a measurement head to obtain the initial geometric parameters of the contact line at the first measurement point of the to-be-measured area; An automatic follow-up continuous measurement step: performing the following processes on the subsequent measurement points: S1. Using the measurement head to collect an image containing the contact line; S2. Segmenting the shape of the contact line in the image and extracting the contact line region; S3. Calculating the pixel center line of the contact line based on the contact line region; S4. Calculating the vertical physical distance between the pixel center line of the contact line and a preset reference pixel point; S5. Calculating the target rotation angle of the measurement head according to the vertical physical distance; S6. Driving the measurement head to rotate to the target rotation angle, measuring and calculating the geometric parameters of the contact line of the current measurement point; S7. Repeating steps S1-S6 to complete the continuous measurement of the geometric parameters of the contact line of the to-be-measured area.

2. The method of claim 1, wherein In step S2, a deep learning segmentation network is used to segment the shape of the contact line in the image.

3. The method of claim 2, wherein the method further comprises: The deep learning segmentation network adopts a Unet network, and its loss function is: Loss=α·BCE+β·DCE+γ·FL, wherein BCE represents the basic binary cross-entropy loss, DCE represents the overlapping loss for the edge region, FL represents the loss for the width of the fuzzy edge region, and α, β, and γ are weight coefficients.

4. The method of claim 1, wherein, In step S3, the pixel center method is used to calculate the pixel center line of the contact line.

5. The method of claim 4, wherein the method further comprises: The pixel center method for calculating the pixel center line of the contact line comprises the following steps: Taking multiple contact net sections at the edges of the contact line region; Determining the left and right edge points of the contact net section, and taking the midpoint of the line connecting the left and right edge points as the center point of the contact net section; Connecting the center points of the multiple contact net sections to form a pixel center line.

6. The method of claim 5, wherein the method further comprises: The pixel center method for calculating the pixel center line of the contact line further comprises the following steps: Using a least squares straight line fitting algorithm to fit the pixel center line: Respectively removing the center points of the contact net sections at a preset proportion before and after the image; Combining the center points of the remaining contact net sections into a point set, and fitting the point set using a straight line equation to obtain the fitted pixel center line.

7. The method of claim 6, wherein the method further comprises: When fitting using the straight line equation, the sum of squares of the vertical distance errors of all data points to the fitted straight line is minimized.

8. The method of claim 1, wherein, The preset reference pixel point is the center point of the image.

9. The method of claim 1, wherein, The target rotation angle is calculated by the following formula: Δa=arctan(H / L)-arctan(H / (L+ΔL)), wherein H is the calculated height of the last measurement point, L is the calculated pull-out value of the last measurement point, and ΔL is the vertical physical distance.

10. A follower-type contact wire geometry measuring device, characterized in that The method comprises the following steps: A walking mechanism for moving on a track; A measurement frame connected to the walking mechanism; A rotatable measurement head mounted on the measurement frame, wherein a camera and a laser range finder are integrated in the rotatable measurement head; A control unit electrically connected with the camera, the laser range finder, and a driving mechanism; the control unit is configured to perform the method of any one of claims 1-9.

Images