Method for efficiently identifying a pendant on a high-speed rail power supply system
By installing recording equipment outside the high-speed train carriages, image acquisition and pixel analysis, combined with slope calculation, can be used to efficiently identify the hanging wires of the high-speed train power supply system. This solves the problem of difficult hanging wire identification in existing technologies, saves labor costs and reduces the misjudgment rate.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHENGZHOU RUHUI INFORMATION TECH CO LTD
- Filing Date
- 2022-01-24
- Publication Date
- 2026-06-26
AI Technical Summary
In the current technology, it is difficult to identify the hanging wires of the power supply equipment along the high-speed railway, the inspection workers have a high labor intensity, and the drone inspection has problems with misjudgment and battery life.
Cameras are installed on the outside of high-speed train carriages. By acquiring and processing images, the initial point of the dropper is identified. The shape of the dropper is determined by pixel analysis and slope calculation. Combined with multi-level sampling loop verification, efficient dropper identification is achieved.
Without the need for drones, drop wire identification can be performed directly outside the high-speed train carriage, saving labor costs, improving identification efficiency, reducing the false judgment rate, and increasing equipment operating speed.
Smart Images
Figure CN116140232B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for testing high-speed railway power supply equipment, and in particular to a method for efficiently identifying the hanging wires of a high-speed railway power supply system. Background Technology
[0002] A power supply system is a system that generates electrical energy and supplies and transmits it to electrical equipment, consisting of a power source system and a power transmission and distribution system. In long-distance power supply systems, there are often problems with line equipment damage and difficult maintenance, such as common issues like bird nests, weathering and breakage of lines, and line wear.
[0003] The dropper is a crucial component of the catenary. The contact wire is suspended from the catenary via the dropper. Adjusting the dropper's length ensures the structural height of the contact suspension and the working height of the contact wire above the rail surface. It increases the suspension points of the contact wire, improving the current-carrying quality of the electric locomotive's pantograph. Its characteristics include high installation precision, high strength, strong current-carrying capacity, corrosion resistance, and long service life. There are various types, including: ordinary link droppers, positioning droppers, flexible cross-span straight droppers, tunnel droppers, windproof droppers, and integral droppers, etc.
[0004] In specific operating conditions, droppers, being linear load-bearing materials, are often more prone to damage than other wiring, and when damage occurs, it is often difficult to detect immediately. Currently, dropper maintenance mainly relies on manual inspections by workers, which are labor-intensive and inefficient. While drones are used for inspections, their limited flight time still presents challenges. Furthermore, drone inspection technology is prone to misinterpretations, often increasing the time workers spend traveling and raising labor costs. Summary of the Invention
[0005] The purpose of this invention is to solve the problems of difficulty in identifying damaged droppers on power supply equipment along high-speed railways and the high labor intensity of inspection workers in the existing technology.
[0006] The specific solution of this invention is:
[0007] A method for efficiently identifying the upper suspension wires of a high-speed rail power supply system is designed, including the following steps:
[0008] (1) Image acquisition: A camera is installed on the outside of the front of the high-speed train carriage. When the high-speed train runs at speeds of 150 km / h or higher, a photograph is taken every 0.2 to 0.5 seconds to form an acquired image. The acquired image is then input into the communication system or the image processing system is input into the timed hard disk to start the recognition and screening.
[0009] (2) Find the suspected initial point of the dropper on the acquired image; starting from the lower right corner of the acquired image and moving towards the upper left corner, check the path pixels horizontally in units of a single pixel. When two or more consecutive dark points are found in a single row, determine that the group of dark points is the suspected initial point group of the dropper, and select the midpoint of the line segment formed by the suspected initial point group of the dropper as the circle point.
[0010] (3) Determine whether a suspected drop string is a secondary sampling range: Take the midpoint of the line segment formed by the initial point group of the suspected drop string in step (2) as the circle point, and define the secondary sampling range with a radius of 2 to 19 pixels. Within the secondary sampling range, check the column of the investigation in step (2) upwards row by row to form dark spot distribution data. Then analyze the data. When the dark spot area of the lower column of the sampling area forms a closed shape, it is determined to be a suspended object in the air. When the dark spot area of the lower column of the sampling area extends to the edge of the secondary sampling range, it is determined to be a suspected drop string.
[0011] (4) For suspected hanging wires, find the distribution of dark line segments of each row of pixels and determine the dark point in the center of the dark line segment: For suspected hanging wires, first extract the intersection points XG1 and XG1' of the dark point area and the secondary sampling range, sample the midpoint XG1' / 2, establish the trend equation from the circle point to the midpoint XG1' / 2, and extract the slope b of the equation. The intersection points XG1 and XG1' are obtained through the following steps (a). Check the pixel points layer by layer from bottom to top and from right to left. Mark the dark point as 0 and the non-dark point as 1. When the scan result is 1…0…1, the 0 point is the circle point. When the scan result is 1…0…, the starting 0 point is XG1. When the scan result first appears 0…0, the ending 0 point is XG1'.
[0012] (5) Determine whether the curve formed by the combination of dark spots forms a hanging string and make graphic annotations: According to the slope generation method in step (4), take the center point of the dark spot group as the center and the distance of 19 to 30 pixels as the radius to establish a four-level sampling ring. First, extract the intersection points XG2 and XG2' of the dark spot area and the secondary sampling range, sample the midpoint XG2' / 2, sample the trend equation between XG2' / 2 and the circle, extract the slope b1 of the trend equation, when b1=b, determine that the area is a hanging string, and make graphic annotations with red boxes. At the same time, provide the geographic coordinates of this image to the background system.
[0013] In specific implementation, according to the required level of the result, it also includes step (6) multi-level verification. The multi-level verification includes using the method of step (5) to gradually establish a multi-level sampling ring with a relative distance of not less than 30 pixels as the radius, and gradually calculate the slope bn of the trend equation of the point area corresponding to each sampling ring, and check it level by level. When there is an inconsistency, the system reminds that there is doubt about the judgment and marks it with a yellow box.
[0014] A screening step is provided between steps (1) and (2). The screening step includes collecting the quadrilateral area S where each power supply column is located, and using S as the image collected in step (2).
[0015] When the tilt is 90 degrees and the bottom edge of the dark area is a horizontal line, it is determined to be a bolt based on the terminal width.
[0016] When the tilt is 90 degrees and the width difference of each layer in the dark area is greater than 1.5 times its own measurement difference, the system will indicate that the judgment is questionable and mark it with a yellow box.
[0017] In practice, it can also be replaced by step (4) using the initial point as the center to determine the secondary sampling range: using the initial point as the center and the distance of a pixels as the radius, define the secondary pre-judgment sampling range, where a = from 2 to 19 pixels. In the pre-judgment sampling range, read the pixel information from bottom to top and from right to left. Assume that the bright area pixel is 1 and the dark area pixel is 0. Form a series of 1 and 0 in each row. When the 0 value of more than three rows in each row increases by a ratio greater than 1.5 and the 0 value in the lower row decreases, it is determined that the sampling range is unreasonable. Adjust the radius to 2a or more and repeat the above steps until there are more than three rows of 0 values with a change value between -1.5 and 1.5. Determine that the sampling range in this step is the secondary sampling range.
[0018] (5) For suspected hanging strings, find the distribution of dark line segments in each row of pixels and determine the dark point in the center of the dark line segment: In the secondary sampling range, read the pixel information. When the scanning result is 1……0……, the point is XG1. When the scanning result first appears 0……0, the end point is XG1'. Connect the two points and take the midpoint of the line segment as the sampling midpoint XG1' / 2. Expand the pixel distance radius again and establish a third-level sampling range. In the third-level sampling range, find the points XG2, XG2', XG2' / 2 and XG1' / 2 and XG2' / 2 in the same way to obtain the trend slope L1: y=b1x+h1, and calculate the slope b1 of the trend slope.
[0019] (6) Determine whether the curve formed by the combination of dark spots forms a hanging string and make graphic annotations: According to the slope generation method in step (4), the center point of the dark spot group is used as the center and the expanded pixel distance is used as the radius to establish a four-level sampling ring to obtain XG3, XG3', XG3' / 2 and L2 and b2. L2: y=b2x+h2. When b1=b2, the area is determined to be a hanging string and is graphicly annotated with a red box. At the same time, the geographical coordinates of this image are provided to the background system.
[0020] The beneficial effects of this invention are as follows:
[0021] A method for efficiently identifying the hanging wires of the high-speed rail power supply system was designed. First, sampling is convenient and does not require the participation of drones. The sampling device can be directly installed outside the high-speed rail carriage. The entire line will be photographed during a single trip. Second, the identified hanging wires are given to the staff. After verification, the staff in the relevant station can be notified immediately for maintenance, avoiding unnecessary detours and saving labor costs.
[0022] Furthermore, there is a complete image screening method, and the method can also be used to reduce the sampling area, saving image processing memory in the image processing background and improving the running speed of the device;
[0023] The detection method, which takes into account the shape characteristics of the suspension string, is highly targeted and has a low false positive rate. Attached Figure Description
[0024] Figure 1 This is an example of an image acquired in step (1) of the present invention;
[0025] Figure 2 This is the book Figure 1 A schematic diagram of another working angle of the middle sampling area A;
[0026] Figure 3 In this invention Figure 2 Schematic diagram of sampling analysis within the secondary sampling area of the district;
[0027] Figure 4 This is a schematic diagram of step (4) in another embodiment;
[0028] Figure 5 This is a schematic diagram of step (5) in another embodiment;
[0029] Figure 6 This is a schematic diagram of step (6) in another embodiment;
[0030] Figure 7 This is a schematic diagram of step (5) in another embodiment.
[0031] 1. Unreasonable sampling range; 2. Reasonable sampling range; O origin. Detailed Implementation
[0032] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention. Example 1
[0033] A method for efficiently identifying the upper suspension wires of a high-speed rail power supply system, see [link to relevant documentation]. Figures 1 to 3 ,
[0034] Includes the following steps:
[0035] (1) Image acquisition: A camera is installed on the outside of the front of the high-speed train carriage. When the high-speed train runs at speeds of 150 km / h or higher, a photograph is taken every 0.2 to 0.5 seconds to form an acquired image. The acquired image is then input into the communication system or the image processing system is input into the timed hard disk to start the recognition and screening.
[0036] (2) Find the suspected initial point of the dropper on the acquired image; starting from the lower right corner of the acquired image and moving towards the upper left corner, check the path pixels horizontally in units of a single pixel. When two or more consecutive dark points are found in a single row, determine that the group of dark points is the suspected initial point group of the dropper, and select the midpoint of the line segment formed by the suspected initial point group of the dropper as the circle point.
[0037] (3) Determine whether a suspected drop string is a secondary sampling range: Take the midpoint of the line segment formed by the initial point group of the suspected drop string in step (2) as the circle point, and define the secondary sampling range with a radius of 2 to 19 pixels. Within the secondary sampling range, check the column of the investigation in step (2) upwards row by row to form dark spot distribution data. Then analyze the data. When the dark spot area of the lower column of the sampling area forms a closed shape, it is determined to be a suspended object in the air. When the dark spot area of the lower column of the sampling area extends to the edge of the secondary sampling range, it is determined to be a suspected drop string.
[0038] (4) For suspected hanging wires, find the distribution of dark line segments of each row of pixels and determine the dark point in the center of the dark line segment: For suspected hanging wires, first extract the intersection points XG1 and XG1' of the dark point area and the secondary sampling range, sample the midpoint XG1' / 2, establish the trend equation from the circle point to the midpoint XG1' / 2, and extract the slope b of the equation. The intersection points XG1 and XG1' are obtained through the following steps (a). Check the pixel points layer by layer from bottom to top and from right to left. Mark the dark point as 0 and the non-dark point as 1. When the scan result is 1…0…1, the 0 point is the circle point. When the scan result is 1…0…, the starting 0 point is XG1. When the scan result first appears 0…0, the ending 0 point is XG1'.
[0039] (5) Determine whether the curve formed by the combination of dark points forms a hanging string and make graphic annotations: According to the slope generation method in step (4), take the center point of the dark point group as the center and establish a four-level sampling ring with a radius of 19 to 30 pixels. First, extract the intersection points XG2 and XG2' of the dark point area and the secondary sampling range, sample the midpoint XG2' / 2, sample the trend equation between XG2' / 2 and the circle, extract the slope b1 of the trend equation, when b1=b, determine that the area is a hanging string, and make graphic annotations with red boxes. At the same time, provide the geographic coordinates of this image to the background system.
[0040] According to the required level of the result, it also includes step (6) multi-level verification. The multi-level verification includes using the method of step (5) to gradually establish a multi-level sampling ring with a relative distance of not less than 30 pixels as the radius, and gradually calculate the slope bn of the trend equation of the point area corresponding to each sampling ring, and check it level by level. When there is an inconsistency, the system will remind that there is doubt about the judgment and mark it with a yellow box.
[0041] A screening step is provided between steps (1) and (2). The screening step includes collecting the quadrilateral area S where each power supply column is located, and using S as the image collected in step (2).
[0042] When the tilt is 90 degrees and the bottom edge of the dark area is a horizontal line, it is determined to be a bolt based on the terminal width.
[0043] When the tilt is 90 degrees and the width difference of each layer in the dark area is greater than 1.5 times its own measurement difference, the shape of the dark area is closer to a sheet than a line. The system will indicate that the judgment is questionable and mark it with a yellow box.
[0044] In this embodiment, when the tilt is 90 degrees and the bottom edge of the dark area is a horizontal line, it is determined to be a bolt based on the terminal width.
[0045] In this application, the suspension wires of the segment have the characteristic that the slope and thickness of the straight line in the dark area are basically the same, that is, the drooping state or the relatively tilted state in the wind. Therefore, the method of calculating the center point of each line segment and comparing the slope is adopted. When the slopes are consistent, and the single end is the endpoint, it can be determined that the point is the suspension wire.
[0046] The algorithm in this application first finds dark spots in the image, and then focuses on scanning around the dark spots. If the area where the dark spot is located is found to be circular or closed, it can be determined that it is not a hanging string. If the overall shape is found to be a slender rod, it can be determined that it is. The method mainly emphasizes measuring whether there is a uniform slope. Due to the small pixel sampling range, the applicant, who has been working on the front line for many years, found that comparing slopes is a relatively simple method. Example 2
[0047] The principle is the same as in Example 1, except that in Example 1, the damage point of the dropper is assumed to be close to a planar cut, and its end is approximately straight or arc-shaped. The center point of its sampling range is determined to be closer to the centerline of the dropper itself. Steps (4) and (5) are replaced with steps (4), (5), and (6), and subsequent steps are increased sequentially.
[0048] The replacement steps include:
[0049] Step (4) Determine the secondary sampling range with the initial point as the center: With the initial point as the center and a pixel distance as the radius, define the secondary pre-judgment sampling range, where a = 2 to 19 pixels. Read pixel information from bottom to top and from right to left within the pre-judgment sampling range. Assume that the bright area pixel is 1 and the dark area pixel is 0. Form a sequence of 1 and 0 in each row. When the 0 values in three or more rows of each row increase by a ratio greater than 1.5 and the 0 values in the lower rows decrease, it is determined that the sampling range is unreasonable. Adjust the radius to 2a or more and repeat the above steps until there are three or more rows of 0 values with a change value between -1.5 and 1.5. Then determine the sampling range within this step as the secondary sampling range.
[0050] (5) For suspected hanging strings, find the distribution of dark line segments in each row of pixels and determine the dark point in the center of the dark line segment: In the secondary sampling range, read the pixel information. When the scanning result is 1……0……, the point is XG1. When the scanning result first appears 0……0, the end point is XG1'. Connect the two points and take the midpoint of the line segment as the sampling midpoint XG1' / 2. Expand the pixel distance radius again and establish a third-level sampling range. In the third-level sampling range, find the points XG2, XG2', XG2' / 2 and XG1' / 2 and XG2' / 2 in the same way to obtain the trend slope L1: y=b1x+h1, and calculate the slope b1 of the trend slope.
[0051] (6) Determine whether the curve formed by the combination of dark spots forms a hanging string and make graphic annotations: According to the slope generation method in step (4), the center point of the dark spot group is used as the center and the expanded pixel distance is used as the radius to establish a four-level sampling ring to obtain XG3, XG3', XG3' / 2 and L2 and b2. L2: y=b2x+h2. When b1=b2, the area is determined to be a hanging string and is graphicly annotated with a red box. At the same time, the geographical coordinates of this image are provided to the background system.
[0052] In this embodiment, considering that the breakage of the suspension string is a tearing fracture, the center point of the sampling range is determined to be at the tip of the tear segment. In this embodiment, the chord length of the arc at the intersection of the sampling range and the dark area is taken, and then the slope of the midpoint line segment between each chord length is estimated. When the slopes are consistent, it can be determined as a damaged suspension string. Example 3
[0053] The principle is the same as in Example 1. The specific difference is that the suspension string in this example is a single shape with a uniform wire diameter. The steps can be simplified as follows: Step 5 (5) For suspected suspension strings, find the distribution of dark line segments of each row of pixels and determine the dark point in the center of the dark line segment: In the secondary sampling range, read the pixel information. When the scanning result is 0……0, 1……1, the point is XG1, and take the other end of the line connecting the 0 points as XG11. The line segment connecting the two points is F1. Expand the pixel distance radius again and establish a third-level sampling range. In the third-level sampling range, find the points XG2 and XG2' in the same way and form the line F2. Compare F2 and F1. If they are of equal length and only the position is offset, they are directly identified as suspension strings.
[0054] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for efficiently identifying the upper suspension wires of a high-speed railway power supply system, characterized in that, The following steps are included: (1) Image acquisition: A camera is installed on the outside of the front of the high-speed train carriage. When the high-speed train runs at speeds of 150 km / h or higher, a photograph is taken every 0.2 to 0.5 seconds to form an image. The image is then input into the communication system or the image processing system is input into the timed hard disk to start the identification and screening. (2) The suspected initial point of the drop wire is found on the image. Starting from the lower right corner of the acquired image and moving towards the upper left corner, check the path pixels horizontally in units of a single pixel. When two or more consecutive dark points are found within a single row, determine that the group of dark points is a suspected initial point group of the drop string. Select the midpoint of the line segment formed by the suspected initial point group of the drop string as the circle point; (3) Determine whether it is a suspected drop string by setting a secondary sampling range: take the midpoint of the line segment formed by the suspected initial point group of the drop string in step (2) as the circle point, and define the secondary sampling range with a radius of 2 to 19 pixels. Within the secondary sampling range, Along the investigation column upwards, investigate row by row to form dark spot distribution data. Then analyze the data. When the dark spot area of the lower investigation column in the sampling area forms a closed shape, it is determined to be a suspended object in the air. When the dark spot area of the lower investigation column in the sampling area extends to the edge of the secondary sampling range, it is determined to be a suspected hanging string. (4) For the suspected hanging string, find the distribution of dark line segments of each row of pixels and determine the dark spot in the center of the dark line segment: For the suspected hanging string, first extract the intersection points XG1 and XG1' of the dark spot area and the secondary sampling range, sample the midpoint XG1' / 2, and build The trend equation from the circle point to the midpoint XG1' / 2 is established, and the slope b of the equation is extracted. The intersection points XG1 and XG1' are obtained through the following steps: (a) Check the pixels layer by layer from bottom to top and from right to left. Mark dark points as 0 and non-dark points as 1. When the scan result is 1...0...1, the 0 point is the circle point. When the scan result is 1...0..., the starting 0 point is marked as XG1. When the scan result first appears 0...0, the ending 0 point is XG1'. (5) Statistically determine whether the curve formed by the combination of dark points is formed. To form a hanging string, perform graphic annotation: Following the slope generation method in step (4), establish a four-level sampling ring with the center point of the dark spot group as the center and a radius of 19 to 30 pixels. First, extract the intersection points XG2 and XG2' of the dark spot area and the secondary sampling range, sample the midpoint XG2' / 2, sample the trend equation between XG2' / 2 and the circle, and extract the slope b1 of the trend equation. When b1=b, determine that the area is a hanging string and mark it with a red box. At the same time, provide the geographic coordinates of this image to the background system.
2. The method for efficiently identifying the upper suspension wire of a high-speed rail power supply system as described in claim 1, characterized in that: According to the required level of the result, it also includes step (6) multi-level verification. The multi-level verification includes using the method of step (5) to gradually establish a multi-level sampling ring with a relative distance of not less than 30 pixels as the radius, and gradually calculate the slope bn of the trend equation of the point area corresponding to each sampling ring, and check it level by level. When there is an inconsistency, the system will remind that there is doubt about the judgment and mark it with a yellow box.
3. The method for efficiently identifying the upper suspension wire of a high-speed rail power supply system as described in claim 1, characterized in that: A screening step is provided between steps (1) and (2). The screening step includes collecting the quadrilateral area S where each power supply column is located, and using S as the image collected in step (2).
4. The method for efficiently identifying the upper suspension wire of a high-speed rail power supply system as described in claim 1, characterized in that: Steps (4) and (5) are replaced as follows: Step (4) Using the initial point as the center, determine the secondary sampling range: Using the initial point as the center and the distance of a pixels as the radius, define the secondary prediction sampling range, where a = 2 to 19 pixels. In the prediction sampling range, read pixel information from bottom to top and from right to left. Assume that the bright area pixel is 1 and the dark area pixel is 0, and form a sequence of 1 and 0 row by row. When the 0 values of three or more rows in each row increase by a ratio greater than 1.5, the results are as follows: When the number of zero values in the square decreases, it is determined that the sampling range is unreasonable. The radius is adjusted to 2a or more, and the above steps are repeated until there are more than 3 rows where the change value of zero value is between -1.5 and 1.
5. The sampling range in this step is determined as the secondary sampling range; (5) For suspected hanging wires, find the distribution of dark line segments of each row of pixels and determine the dark point in the center of the dark line segment: In the secondary sampling range, read the pixel information. When the scanning result is 1......0......, start 0 The point is marked as XG1. When the scan result first appears 0...0, the end point 0 is the line connecting the two points XG1'. The midpoint of the line segment is taken as the sampling midpoint XG1' / 2. The pixel distance radius is expanded again to establish a three-level sampling range. Within the three-level sampling range, the points XG2, XG2', and XG2' / 2 are found in the same way. The line connecting XG1' / 2 and XG2' / 2 is obtained to obtain the trend slope L1: y=b1x+h1. The slope b1 of the trend slope is calculated. (6) To determine whether a hanging string is formed by the curve formed by the combination of dark points, the following graphic annotation is performed: According to the slope generation method in step (4), the center point of the dark point group is used as the center and the expanded pixel distance is used as the radius to establish a four-level sampling ring to obtain XG3, XG3', XG3' / 2 and L2 and b2. L2: y=b2x+h2. When b1=b2, the area is determined to be a hanging string. The graphic annotation is performed with a red box, and the geographical coordinates of the image are provided to the background system.