Rapid defect positioning method for high-speed railway steel rail inspection

Abstract

The invention belongs to the technical field of defect detection, and discloses a rapid defect positioning method for high-speed railway steel rail inspection, which comprises the following steps: a line laser scanner synchronously moves along with a track inspection trolley and scans a steel rail tread to obtain a steel rail tread point cloud; the method comprises the following steps: separating n line point clouds which are parallel to a steel rail direction and have the same spacing from a steel rail tread point cloud on the basis of defect detection precision requirements, and quickly and preliminarily positioning defects according to the difference of points on the line point clouds; extracting point clouds of defects and surrounding areas from the steel rail tread point clouds by utilizing the preliminarily determined defect positions, and performing clustering processing to obtain defect area point clouds; and performing coordinate system conversion on the defect area point cloud to obtain coordinates of each point in the point cloud in a world coordinate system so as to complete defect positioning. According to the invention, the defect can be quickly positioned in the inspection process of the high-speed railway steel rail.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and particularly relates to a method for quickly locating defects in the inspection of high-speed railway rails. Background Art

[0002] The tread surface of high-speed railway rails is in long-term contact and friction with the wheels, which easily leads to the formation of defects such as spalling and depression, threatening the driving safety of high-speed trains. In order to repair the defects formed on the tread surface of high-speed railway rails in a timely manner, it is necessary to quickly determine the specific location of the defects. During the inspection of high-speed railway rails, a scanner is often used to continuously scan the tread surface of the rails, generating a large amount of point cloud data. By processing the point cloud data, the location of the defects can be achieved. However, the conventional defect location method based on point cloud requires clustering processing of the large amount of point cloud data of the rail tread surface, extracting the point cloud data of the defect area, and finally determining the location of the defect. This process consumes a large amount of computing resources and seriously reduces the location efficiency of the rail tread surface defects. Therefore, there is an urgent need to propose a method that can quickly locate defects and ensure accuracy during the inspection of high-speed railways. Summary of the Invention

[0003] In order to solve the above problems, the present invention proposes a method for quickly locating defects in the inspection of high-speed railway rails, including the following steps:

[0004] S1, the line laser scanner moves synchronously with the track inspection trolley and scans the tread surface of the rail to obtain the point cloud of the rail tread;

[0005] S2, based on the defect detection accuracy requirement, n line point clouds parallel to the rail direction and with the same spacing are separated from the point cloud of the rail tread, and the quick preliminary location of the defect is realized according to the differences of the points on the line point clouds;

[0006] S3, using the preliminarily determined defect location, the point cloud of the defect and its surrounding area is extracted from the point cloud of the rail tread, and after clustering processing, the point cloud of the defect area is obtained;

[0007] S4, the coordinate system of the point cloud of the defect area is transformed to obtain the coordinates of each point in the world coordinate system in this point cloud, and the location of the defect is completed.

[0008] Further, in S1, the line laser scanner is installed on the front side of the track inspection trolley. The line laser it emits is parallel to the horizontal plane and perpendicular to the rail direction. During the inspection process, the line laser scanner moves synchronously along the rail direction with the track inspection trolley at a constant speed, emits line laser at a constant frequency to scan the rail tread; after the scanning is completed, the obtained point cloud is filtered using the radius filtering algorithm to obtain the rail tread point cloud; the rail tread point cloud is located in the coordinate system of the line laser scanner. This coordinate system takes the direction perpendicular to the rail direction and the gravity direction as the X-axis direction, the direction parallel to the rail direction as the Y-axis direction, the opposite direction of gravity as the Z-axis direction, and the origin is located at the position of the line laser scanner before the inspection starts.

[0009] Further, in S2, n line point clouds parallel to the Y-axis direction and with the same spacing are separated from the rail tread point cloud. The n line point clouds equally divide the width d of the rail tread point cloud into n + 1 parts. If the accuracy requirement for defect detection is to be able to detect defects with a width greater than or equal to a, then it can be obtained that:

[0010]

[0011] Furthermore, the calculation formula for n can be obtained as:

[0012]

[0013] Traverse each point in the n line point clouds, calculate the average height difference between the i-th point and its m neighboring points in each line point cloud. If the calculated value of a certain point is greater than the set threshold, then it is determined that this point belongs to the defect area. Screen out all the points belonging to the defect area and calculate the minimum value y of the y coordinate values among them min and the maximum value y max , to achieve a quick preliminary positioning of the defect.

[0014] Further, in S3, all the points with y coordinate values within [y min , y max are extracted from the rail tread point cloud to obtain the point cloud of the defect and its surrounding area; calculate the curvature and normal vector of each point in the point cloud of the defect and its surrounding area, and use the region growing algorithm to perform clustering processing on this point cloud to obtain the point cloud of the defect area.

[0015] Further, in S4, a world coordinate system is established with the direction parallel to the rail as the X'-axis direction, the direction perpendicular to the rail direction and the gravity direction as the Y'-axis direction, and the opposite direction of gravity as the Z'-axis direction. The origin of this coordinate system is located at the position where the line laser scanner irradiates the rail tread before the inspection starts. Respectively obtain the line point cloud in the line laser scanner coordinate system and the line point cloud in the world coordinate system at the position where the line laser scanner irradiates the rail tread before the inspection starts. Perform registration processing on the above two line point clouds to calculate the rotation matrix and the translation matrix. Use the rotation matrix and the translation matrix to transform the point cloud of the defect area from the line laser scanner coordinate system to the world coordinate system, obtain the coordinates of each point in the point cloud of the defect area in the world coordinate system, and complete the precise positioning of the defect.

[0016] Compared with the prior art, the above technical solution conceived by the present invention mainly has the following beneficial effects:

[0017] A method for rapid defect positioning for high-speed railway rail inspection proposed by the present invention, on the basis of considering the requirements for defect detection accuracy, separates n line point clouds parallel to the rail direction and with the same spacing from the point cloud on the rail tread, and realizes the rapid positioning of defects by analyzing the differences of each point in the line point cloud, avoiding directly clustering the massive point cloud data on the rail tread, and can significantly improve the efficiency of defect positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flowchart of a method for rapid defect positioning for high-speed railway rail inspection according to the present invention;

[0019] Figure 2 is a schematic diagram of a line laser scanner scanning the rail tread;

[0020] Figure 3 is a schematic diagram of the point cloud on the rail tread;

[0021] Figure 4 is a schematic diagram of the position for separating line point clouds from the point cloud on the rail tread;

[0022] Figure 5 is a schematic diagram of three separated line point clouds;

[0023] Figure 6 is a schematic diagram of the point cloud of a depression defect and its surrounding area;

[0024] Figure 7 is a schematic diagram of the point cloud of the depression defect area;

[0025] Figure 8 is a schematic diagram of the world coordinate system;

[0026] Figure 9 is a schematic diagram of the line point cloud in the line laser scanner coordinate system;

[0027] Figure 10 It is a schematic diagram of line point cloud in the world coordinate system;

[0028] Figure 11 It is a schematic diagram of point cloud of the sunken defect area in the world coordinate system;

[0029] In the figure: 1 - line laser scanner, 2 - rail, 3 - rail tread, 4 - sunken defect, 5 - line laser, 6 - the first line point cloud, 7 - the second line point cloud, 8 - the third line point cloud, 9 - line point cloud in the line laser scanner coordinate system, 10 - line point cloud in the world coordinate system. Specific implementation mode

[0030] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the drawings and specific embodiments. The schematic implementation mode of the present invention is only used to explain the present invention and is not a limitation to the present invention.

[0031] Refer to Figures 1-11 , this embodiment provides a method for quickly locating defects for high - speed railway rail inspection, including the following steps:

[0032] S1, the line laser scanner 1 moves synchronously with the track inspection trolley and scans the rail tread 3 to obtain the point cloud of the rail tread 3.

[0033] In this embodiment, the line laser scanner 1 is installed in front of the side of the track inspection trolley and emits a line laser 5 parallel to the horizontal plane and perpendicular to the direction of the rail 2; the frequency of the line laser scanner 1 emitting the line laser 5 is 3000Hz, the speed of the track inspection trolley is 1m / s, and the line laser scanner 1 moves synchronously with the track inspection trolley along the direction of the rail 2 and scans the rail tread 3, as Figure 2 shown. After the scanning is completed, the obtained point cloud is filtered by using the radius filtering algorithm to obtain the point cloud of the rail tread 3, as Figure 3 shown, specifically including:

[0034] Traverse each point in the obtained point cloud. If the number of points within the neighborhood radius of a certain point is less than the set threshold, then this point is regarded as an outlier and filtered out until all outliers are filtered out.

[0035] The point cloud of the rail tread 3 is located in the coordinate system of the line laser scanner 1. This coordinate system takes the direction perpendicular to the direction of the rail 2 and the gravity direction as the X - axis direction, the direction parallel to the rail 2 as the Y - axis direction, and the opposite direction of gravity as the Z - axis direction. The origin is located at the position of the line laser scanner before the start of the inspection.

[0036] S2. Based on the defect detection accuracy requirement, n line point clouds parallel to the direction of the rail 2 with the same spacing are separated from the point cloud of the rail tread 3. According to the differences of the points on the line point clouds, a rapid preliminary positioning of the defect is realized.

[0037] In this embodiment, n line point clouds parallel to the Y-axis direction with the same spacing are separated from the point cloud of the rail tread 3. The n line point clouds equally divide the width d of the point cloud of the rail tread 3 into n + 1 parts. If the defect detection accuracy requirement is to be able to detect defects with a width greater than or equal to a, then it can be obtained that:

[0038]

[0039] Furthermore, the calculation formula for n can be obtained as:

[0040]

[0041] The width of the point cloud of the rail tread 3 is 46 mm, and the defect detection accuracy requirement is to be able to detect a concave defect 4 with a width greater than or equal to 12 mm. Then it can be obtained that three line point clouds need to be separated from the point cloud of the rail tread 3. The positions of the first line point cloud 6, the second line point cloud 7, and the third line point cloud 8 in the point cloud of the rail tread 3 are as Figure 4 shown, and the three separated line point clouds are as Figure 5 shown.

[0042] Traverse each point in the three line point clouds, calculate the average height difference between the i-th point and its m neighboring points in each line point cloud. If the calculated value of a certain point is greater than the set threshold, then it is determined that this point belongs to the area of the concave defect 4. Screen out all the points belonging to the area of the concave defect 4, and calculate the minimum value y of the y coordinate values among them min is 83 mm, and the maximum value y of the y coordinate values max is 97 mm, realizing the rapid preliminary positioning of the concave defect 4.

[0043] S3. Using the preliminarily determined defect position, extract the point cloud of the defect and its surrounding area from the point cloud of the rail tread 3, and perform clustering processing to obtain the point cloud of the defect area.

[0044] In this embodiment, all the points with y coordinate values within [y min , y max are extracted from the point cloud of the rail tread 3 to obtain the point cloud of the concave defect 4 and its surrounding area, as Figure 6 shown. Calculate the curvature and normal vector of each point in the point cloud of the concave defect 4 and its surrounding area, and use the region growing algorithm to perform clustering processing on this point cloud to obtain the point cloud of the concave defect 4 area, as Figure 7 shown, specifically including:

[0045] Select a point in the point cloud of the depression defect 4 and its surrounding area as the seed point. Starting from the seed point, according to the preset similarity thresholds of curvature and normal vector, gradually add the points in the neighborhood that meet the similarity thresholds to the growing area. When there are no points in the neighborhood that meet the similarity thresholds, then use other points in the neighborhood as the seed points to continue growing until the clustering process of the point cloud of the depression defect 4 and its surrounding area is completed.

[0046] S4. Perform coordinate system transformation on the point cloud of the defect area to obtain the coordinates of each point in the point cloud in the world coordinate system, and complete the positioning of the defect.

[0047] In this embodiment, take the direction parallel to the rail 2 as the X'-axis direction, the direction perpendicular to the rail 2 and the gravity direction as the Y'-axis direction, and the opposite direction of gravity as the Z'-axis direction to establish a world coordinate system. The origin of this coordinate system is located at the position where the line laser scanner 1 irradiates on the rail tread 3 before the inspection starts, as Figure 8 shown.

[0048] Obtain the line point cloud 9 in the line laser scanner coordinate system at the position where the line laser scanner 1 irradiates on the rail tread 3 before the inspection starts, as Figure 9 shown; obtain the line point cloud 10 in the world coordinate system at the position where the line laser scanner 1 irradiates on the rail tread 3 before the inspection starts, as Figure 10 shown.

[0049] First, use the sampling consistency initial registration algorithm to perform rough registration on the line point cloud 9 in the line laser scanner coordinate system and the line point cloud 10 in the world coordinate system, and calculate the rotation matrix R0 and translation matrix T0 in the rough registration stage. Specifically, it includes:

[0050] Calculate the fast point feature histogram descriptors of each point in the two line point clouds, select the points with similar fast point feature histogram descriptors in the two line point clouds as corresponding points to form a corresponding point set, select some corresponding points from the corresponding point set to solve the rotation matrix and translation matrix, and calculate the corresponding registration error. After multiple solutions, select the rotation matrix and translation matrix with the smallest registration error as the rotation matrix R0 and translation matrix T0 in the rough registration stage.

[0051] Then, use the iterative closest point algorithm to perform fine registration on the two line point clouds, and calculate the rotation matrix R1 and translation matrix T1 in the fine registration stage. Specifically, it includes:

[0052] Calculate the Euclidean distance between each point in the two line point clouds, select the points with smaller Euclidean distance as corresponding points to form a corresponding point set, use the corresponding point set to iteratively solve the rotation matrix and translation matrix, calculate the corresponding registration error, until the registration error is lower than the set threshold, end the iterative process, and use the current rotation matrix and translation matrix as the rotation matrix R1 and translation matrix T1 in the fine registration stage.

[0053] Multiply the coordinates of each point in the point cloud of the sunken defect 4 area by the rotation matrix R0 in the rough registration stage and add the translation matrix T0, and then multiply by the rotation matrix R1 in the fine registration stage and add the translation matrix T1, so that the point cloud of the sunken defect 4 area in the coordinate system of the line laser scanner 1 is transformed into the world coordinate system, as Figure 11 shown. Obtain the coordinates of each point in the point cloud of the sunken defect 4 area in the world coordinate system, and complete the precise positioning of the sunken defect 4.

[0054] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for rapid defect location in the inspection of high-speed railway rails, characterized in that, It includes the following steps: S1. The line laser scanner moves synchronously with the track inspection trolley and scans the rail tread surface to obtain the rail tread point cloud; S2. Based on the defect detection accuracy requirements, n line point clouds parallel to the rail direction and with the same spacing are separated from the rail tread point cloud, and the rapid preliminary positioning of defects is realized according to the differences of the points on the line point clouds; S3. The point cloud of the defect and its surrounding area is extracted from the rail tread point cloud using the preliminarily determined defect position, and the defect area point cloud is obtained after clustering processing; S4. The coordinate system of the defect area point cloud is transformed to obtain the coordinates of each point in the world coordinate system in this point cloud, and the positioning of the defect is completed.

2. The rapid defect location method for high-speed railway rail inspection according to claim 1, wherein, In S1, the line laser scanner is installed in front of the side of the track inspection trolley. The line laser it emits is parallel to the horizontal plane and perpendicular to the rail direction. During the inspection process, the line laser scanner moves synchronously along the rail direction with the track inspection trolley with a constant speed, emits the line laser at a constant frequency, and scans the rail tread surface; after the scanning is completed, the obtained point cloud is filtered using the radius filtering algorithm to obtain the rail tread point cloud; the rail tread point cloud is located in the line laser scanner coordinate system. This coordinate system takes the direction perpendicular to the rail direction and the gravity direction as the X-axis direction, the direction parallel to the rail direction as the Y-axis direction, and the opposite direction of the gravity as the Z-axis direction, and the origin is located at the position of the line laser scanner before the start of the inspection.

3. A method for rapid defect location in high-speed railway rail inspection according to claim 1, characterized in that In S2, n line point clouds parallel to the Y-axis direction and with the same spacing are separated from the rail tread point cloud. The n line point clouds equally divide the width d of the rail tread point cloud into n + 1 parts. If the defect detection accuracy requirement is to be able to detect defects with a width greater than or equal to a, then it can be obtained that: Furthermore, the calculation formula for n can be obtained as: Traverse each point in the n-line point cloud, calculate the average height difference between the i-th point and its m neighboring points in each line point cloud. If the calculated value of a certain point is greater than the set threshold, then determine that this point belongs to the defect area. Screen out all points belonging to the defect area and calculate the minimum value y of the y coordinate values among them min and the maximum value y max , to achieve a rapid preliminary positioning of the defect.

4. A method for rapid defect location in high - speed railway rail inspection according to claim 1, characterized in that, In S3, all points with y coordinate values within [y min , y max are extracted from the rail tread point cloud to obtain the point cloud of the defect and its surrounding area; Calculate the curvature and normal vector of each point in the point cloud of the defect and its surrounding area, and use the region growing algorithm to perform clustering processing on this point cloud to obtain the defect area point cloud.

5. A method for rapid defect location in high-speed railway rail inspection according to claim 1, characterized in that, In S4, a world coordinate system is established with the direction parallel to the rail direction as the X'-axis direction, the direction perpendicular to the rail direction and the gravity direction as the Y'-axis direction, and the opposite direction of the gravity as the Z'-axis direction. The origin of this coordinate system is located at the position where the line laser scanner irradiates the rail tread before the start of the inspection; respectively obtain the line point cloud in the line laser scanner coordinate system and the line point cloud in the world coordinate system at the position where the line laser scanner irradiates the rail tread before the start of the inspection; perform registration processing on the above two line point clouds to calculate the rotation matrix and the translation matrix; use the rotation matrix and the translation matrix to transform the defect area point cloud from the line laser scanner coordinate system to the world coordinate system to obtain the coordinates of each point in the defect area point cloud in the world coordinate system, and complete the precise positioning of the defect.

Images