Steel rail inclined crack detection and rating method based on machine vision and nondestructive inspection

Abstract

The invention provides a steel rail inclined crack detection and rating method based on machine vision and nondestructive inspection, and belongs to the technical field of steel rail maintenance. The method specifically comprises the following steps: performing image acquisition on a rail surface state of a steel rail, and acquiring damage depth data of the steel rail; steel rail oblique-line-shaped stripping crack recognition is conducted on the collected images based on machine vision, and the morphological characteristics of oblique-line-shaped stripping cracks are extracted; and performing grade division based on different development stage characteristics on the steel rail oblique-line-shaped stripping cracks in combination with the morphological characteristics and the damage depth data. According to the rating method provided by the invention, a machine vision technology and a multi-dimensional data fusion detection method are utilized, so that the long-mileage steel rail can be quickly and efficiently detected, the workload and time cost of manual inspection are reduced, the detection efficiency is improved, and the high-efficiency requirement of high-speed railway steel rail detection is met.

Description

Technical Field

[0001] The present invention relates to the technical field of rail repair, and in particular to a method for detecting and grading rail oblique cracks based on machine vision and non-destructive testing. Background Art

[0002] Oblique peeling cracks in rails (abbreviated as "oblique cracks") are difficult to detect in the early stages, develop rapidly in the middle stages, and are difficult to repair in the late stages. They are often found on straight sections or large-radius curves on passenger or passenger-freight railways, and are very harmful.

[0003] Diagonal crack damage is highly concealed in its early stages, typically initiating at the edge of the rail's smooth band before any changes to the rail's cross-sectional or longitudinal profiles. This typically does not impact train operations and is difficult to detect through train operation responses. As the crack rapidly propagates toward the center of the smooth band, it expands internally below the rail surface, widening the smooth band. After a large area of ​​expansion develops within the rail, significant rail collapse occurs, eventually developing into large-scale debonding. Depressions and dark spots gradually appear on the rail surface, and the crack depth can reach over 3mm. In severe cases, it can develop into a core damage or even rail breakage, making repair difficult and posing a significant threat to high-speed rail safety. Diagonal crack damage can be eliminated by grinding in the early stages, but in the middle or late stages, the difficulty and cost of repair increases significantly. Discovery of late-stage diagonal debonding cracks necessitates costly replacement of the entire rail.

[0004] While scholars both domestically and internationally have conducted extensive research on the formation mechanisms and characteristics of diagonal rail peeling cracks, there has been no research on rapid detection methods for these cracks during in-service operation, nor on the assessment and rating of these cracks at different stages. Currently, the lack of a clear understanding of the damage characteristics of diagonal cracks at different stages of development presents significant challenges for maintenance and repair, making it difficult to fully eliminate this potential threat within a short period of time.

[0005] Specifically, existing rail damage detection methods are not accurate enough in determining the extent of diagonal crack damage. In particular, there is a lack of clear standards and guidance for categorizing diagonal crack grades and development stages. This lack of clarity makes it difficult for on-site maintenance personnel to accurately determine the extent of rail damage, hindering decision-making regarding rail replacement, repair, and maintenance.

[0006] Manual inspection relies on the experience and skills of the inspector, is time-consuming, and is susceptible to human factors, resulting in low efficiency. Furthermore, manual inspection typically requires a significant amount of human resources, resulting in high labor and energy costs. It also struggles to meet the large-scale, high-frequency inspection requirements required for efficient and accurate detection of diagonal cracks in high-speed rails.

[0007] While manual single-track flaw detectors, manual dual-track flaw detectors, and large-scale flaw detection vehicle inspection technologies offer significant efficiency advantages and can quickly cover a wide inspection area, their ability to identify subtle defects is limited. This is especially true for diagonal cracks of Grade III or below, which are often difficult to accurately identify using existing inspection technologies. These defects are often subtle, and the complex morphology of diagonal cracks leads to a high rate of missed detection of early-stage diagonal cracks.

[0008] Existing rail flaw detection technologies typically rely on a single type of data (ultrasonic signals), limiting the comprehensiveness of test results. Single-dimensional data cannot fully reflect the true condition of the inspected object, especially when dealing with complex damage or multiple factors. The limitations of a single data source are even more pronounced. Relying solely on ultrasonic testing can overlook subtle surface cracks. The lack of integrated analysis of multi-dimensional data makes existing technologies inadequate for damage assessment, making it difficult to provide comprehensive and accurate test results.

[0009] In view of this, the inventor, based on many years of production design experience in this field and related fields, has designed a rail oblique crack detection and rating method based on machine vision and non-destructive testing after repeated experiments, in order to solve the problems existing in the prior art. Summary of the Invention

[0010] The purpose of the present invention is to provide a rail oblique crack detection and rating method based on machine vision and non-destructive testing, which can detect early rail oblique peeling crack damage and perform grade assessment on the discovered oblique peeling crack damage.

[0011] To achieve the above-mentioned purpose, the present invention proposes a method for detecting and rating oblique cracks in rails based on machine vision and non-destructive testing, wherein images of the rail surface state are collected and non-destructive testing is performed on the damage depth data of the rails;

[0012] Based on machine vision, the rail oblique peeling cracks are identified from the acquired images and the morphological features of the oblique peeling cracks are extracted;

[0013] Combining the morphological characteristics and damage depth data of non-destructive testing, the rail oblique peeling cracks are classified into different levels based on the characteristics of different development stages.

[0014] Compared with the prior art, the present invention has the following characteristics and advantages:

[0015] This paper proposes a method for detecting and grading oblique cracks in rails based on machine vision and nondestructive testing. By integrating 2D and 3D image information with nondestructive testing data such as eddy current, magnetic flux leakage, and ultrasonic waves, it can more comprehensively and accurately identify the early damage characteristics of oblique linear peeling cracks in rails. This effectively addresses the difficulty of detecting early oblique cracks in existing technologies and improves detection reliability and accuracy. This detection method, which utilizes machine vision technology and multi-dimensional data fusion, can quickly and efficiently inspect long mileage rails, reducing the workload and time cost of manual inspections, improving detection efficiency, and meeting the high-efficiency requirements of high-speed rail inspections.

[0016] The present invention proposes a method for detecting and rating oblique cracks in rails based on machine vision and nondestructive testing. Based on image features and nondestructive testing signal features, it can clearly define the different development stages and damage degrees of oblique cracks, provide maintenance personnel with clear damage rating standards, and help them formulate maintenance strategies more scientifically, avoid excessive or insufficient maintenance, and improve the scientificity and rationality of maintenance decisions. Early detection of oblique crack damage allows rail grinding and maintenance to be intervened in time, avoiding further crack expansion that leads to rail scrapping, reducing the number of rail replacements, and thus reducing the economic cost of rail maintenance. At the same time, accurate damage assessment and maintenance decisions also help optimize the allocation of maintenance resources and further reduce maintenance costs. Timely detection and treatment of early oblique crack damage can effectively prevent serious accidents such as rail breakage caused by crack expansion, ensure the safety of train operation, reduce traffic safety risks caused by rail damage, and improve the overall safety and reliability of rail transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are for illustrative purposes only and are not intended to limit the scope of the present invention in any way. In addition, the shapes and proportional dimensions of the various components in the drawings are merely illustrative and are used to help understand the present invention, and are not intended to specifically limit the shapes and proportional dimensions of the various components of the present invention. Those skilled in the art can select various possible shapes and proportional dimensions to implement the present invention according to specific circumstances under the guidance of the present invention.

[0018] Figure 1 Schematic diagram of the rating method of the present invention;

[0019] Figure 2 It is a schematic diagram of the rating process of the present invention;

[0020] Figure 3 This is a schematic diagram of the process of block location and size measurement of the present invention;

[0021] Figure 4 This is a schematic diagram of the three-dimensional point cloud contour image of the block drop of the present invention;

[0022] Figure 5This is an example diagram of the identification mark of the present invention;

[0023] Figure 6-1 Schematic diagram of micro cracks on the rail surface of the present invention;

[0024] Figure 6-2 This is a schematic diagram of a level I oblique crack on the rail surface of the present invention;

[0025] Figure 6-3 This is a schematic diagram of a level II oblique crack on the rail surface of the present invention;

[0026] Figure 6-4 This is a schematic diagram of a level III oblique crack on the rail surface of the present invention;

[0027] Figure 6-5 This is a schematic diagram of the rail surface oblique cracks and chipping at level I of the present invention;

[0028] Figure 6-6 This is a schematic diagram of the rail surface oblique cracks and falling pieces rating level II of the present invention;

[0029] Figure 6-7 This is a schematic diagram of the grade III rating of the rail surface with oblique cracks and falling pieces according to the present invention. DETAILED DESCRIPTION

[0030] The details of the present invention can be more clearly understood by referring to the accompanying drawings and the description of the specific embodiments of the present invention. However, the specific embodiments of the present invention described herein are only for the purpose of explaining the present invention and are not to be construed as limiting the present invention in any way. Based on the teachings of the present invention, a skilled person can conceive of any possible variations based on the present invention, and such variations should be considered to fall within the scope of the present invention.

[0031] like Figure 1 and Figure 2 As shown, the present invention proposes a method for detecting and grading oblique cracks in rails based on machine vision and non-destructive testing, wherein images of the rail surface condition are collected, and non-destructive testing is performed on the damage depth data of the rails; based on machine vision, the collected images are used to identify oblique peeling cracks in the rails and extract the morphological features of the oblique peeling cracks; and the oblique peeling cracks in the rails are graded based on the characteristics of different development stages by combining the morphological features and the damage depth data obtained by non-destructive testing.

[0032] The present invention proposes a method for detecting and grading oblique cracks in rails based on machine vision and non-destructive testing. By integrating 2D and 3D image information and non-destructive testing data such as eddy current, magnetic flux leakage, and ultrasonic waves, it can more comprehensively and accurately identify the early damage characteristics of oblique linear peeling cracks in rails, thereby improving the reliability and accuracy of detection. At the same time, the detection method using machine vision technology and multi-dimensional data fusion can quickly and efficiently detect long-mileage rails, reduce the workload and time cost of manual inspections, and improve detection efficiency. In addition, the oblique crack grading method based on image features and non-destructive testing signal features can clearly define the different development stages and damage degrees of oblique cracks, provide maintenance personnel with clear damage rating standards, help them formulate maintenance strategies more scientifically, avoid excessive or insufficient maintenance, improve the scientificity and rationality of maintenance decisions, reduce the economic cost of rail maintenance, ensure the safety of train operation, and improve the overall safety and reliability of rail transit.

[0033] In an optional embodiment of the present invention, the acquired image is a 2D image and / or a 3D image.

[0034] Specifically, the acquisition of image information of oblique crack damage includes 2D image information acquisition and / or 3D image information acquisition. Among them, 2D image information acquisition uses line scanning imaging technology for the inspection of rail surface surface defects, which continuously and uninterruptedly images the rail surface status and collects and stores images of oblique crack damage on the rail surface. 3D image information acquisition uses a 3D vision system to use geometric or physical principles and non-contact optical imaging technology to identify the spatial position of the object being measured. The object being measured will convert a 3D point cloud data set to quantitatively measure the accurate numerical information of the defect and realize the quantification of the defect in the depth direction. By acquiring 2D and / or 3D images, the accuracy and reliability of the test results are significantly improved. 2D images can clearly present the rail surface status and provide information on the morphology and distribution of cracks; 3D images provide crack depth data through depth measurement, providing a more comprehensive reference for crack rating.

[0035] The data collection requirements are:

[0036] 1. 2D rail surface image acquisition requires effective pixels to clearly capture the entire rail head cross-section. Vertical resolution of rail surface image acquisition should be ≤ 0.6mm, and lateral resolution should be ≤ 0.1mm. Each captured photo must fully display at least one severe diagonal crack. If a 3D camera is also available, the 3D rail surface image must clearly cover the entire width of the rail head cross-section, and each image segment must fully display at least one severe diagonal crack. The accuracy must reach ±0.1mm.

[0037] 2. The data contains mileage information and can be replayed and viewed frame by frame.

[0038] In an optional embodiment of the present invention, damage depth data is collected by ultrasonic testing, magnetic flux leakage testing or eddy current testing. Non-destructive testing (NDT) is a method of inspecting the quality, integrity or performance of an object without destroying its structure or function. Common non-destructive testing methods include: ultrasonic testing (using sound waves to detect internal defects), radiographic testing (using X-ray or gamma-ray imaging to detect defects), eddy current testing (based on electromagnetic induction to detect surface and near-surface defects of conductive materials) and magnetic flux leakage testing (detecting defects in ferromagnetic materials through changes in magnetic fields), etc. Each method is suitable for the detection of different materials and defect types.

[0039] Specifically, nondestructive testing for diagonal crack damage involves simultaneously acquiring damage depth data across the entire rail surface and the wheel-rail contact area using nondestructive testing methods such as ultrasonic testing, magnetic flux leakage testing, radiographic testing, or eddy current testing. This data is used to determine the damage grade of diagonal cracks and is subsequently used for multi-data analysis and verification and annotation of diagonal crack data to identify and address gaps. These nondestructive testing methods effectively identify crack depth within the rail, providing highly accurate test data and ensuring the accuracy and reliability of test results. Using deep learning and image recognition processing, the morphological characteristics and damage severity of diagonal cracks of varying grades are quantitatively analyzed, significantly improving the accuracy and reliability of test results. Ultrasonic testing, magnetic flux leakage testing, radiographic testing, and eddy current testing each offer unique advantages, enabling them to capture internal rail damage information from diverse angles and dimensions, ensuring comprehensive and accurate test results. This multi-dimensional data collection approach not only improves test accuracy but also enhances the credibility of test results, providing a solid data foundation for rail maintenance decisions.

[0040] In an optional example of this embodiment, the damage depth data collection covers the entire wheel-rail contact area of ​​the transverse rail surface.

[0041] Specifically, the requirements for collecting information data for nondestructive testing of oblique crack damage include:

[0042] 1. Coverage requirements: Since the oblique cracks extend below the rail surface, they can basically cover the entire wheel-rail contact area of ​​the rail surface. Therefore, the non-destructive testing is required to cover the entire wheel-rail contact area on the transverse rail surface so as to find the deepest extension position below the rail surface.

[0043] 2. Resolution and accuracy requirements: Detection accuracy must reach the sub-millimeter level (0.02mm level) to ensure accurate identification of tiny oblique cracks.

[0044] 3. The data contains mileage information, which corresponds well to the mileage information of 2D and / or 3D images and can be replayed and viewed at different locations.

[0045] Using rapid rail inspection equipment equipped with 2D and multi-channel eddy current flaw detectors, or 2D+3D cameras and multi-channel eddy current flaw detectors, rail surface images are captured, along with damage depth data. Machine vision-based data analysis software is used to identify and mark diagonal rail debonding cracks. Eddy current flaw detection data and rail surface image features are then combined to classify these cracks into different development stages. Nondestructive testing methods such as eddy current, magnetic flux leakage, and ultrasonic testing simultaneously acquire damage depth data across the entire rail surface in the wheel-rail contact area. This data is used to determine the damage grade of diagonal cracks and is subsequently used for multi-data analysis and verification of annotated diagonal crack data to identify and address gaps. Deep learning, image recognition, and nondestructive testing are used to quantify the morphological characteristics and damage severity of diagonal cracks of different grades, significantly improving detection results and ensuring early detection and accurate assessment of diagonal rail debonding cracks. This effectively addresses the existing challenges in detecting early diagonal cracks.

[0046] In an optional embodiment of the present invention, the identification results of the oblique linear peeling cracks on the rail are oblique crack level I, oblique crack level II, oblique crack level III, oblique crack and chipping level I, oblique crack and chipping level II and oblique crack and chipping level III.

[0047] Specifically, the oblique peeling cracks on the rails are identified and marked through machine vision data analysis software. The eddy current flaw detection data and rail surface image features are further combined to classify the oblique peeling cracks on the rails into different levels based on the characteristics of different development stages. The identification results include oblique crack level I, oblique crack level II, oblique crack level III, oblique crack and chipping level I, oblique crack and chipping level II, and oblique crack and chipping level III. Among them, oblique crack levels I to III refer to the situation where oblique cracks appear on the surface of the rail but do not reach the point of chipping, and oblique crack and chipping levels I to III refer to the state where the expansion of oblique cracks leads to local metal peeling. Through deep learning and image recognition processing methods, multi-dimensional data is integrated to quantitatively analyze the damage morphology characteristics and damage degree of oblique cracks of different levels, which can more accurately identify the different damage states of oblique peeling cracks on the rails, effectively solving the problem of inaccurate identification of early oblique crack damage in the existing technology.

[0048] In an optional example of this embodiment, the identified rail oblique peeling crack is marked, such as Figure 4The figure below shows a 3D point cloud outline image of a broken rail after marking. The crack location and severity information marked in the 3D point cloud outline image can be clearly displayed to the user, providing an intuitive reference for maintenance personnel. This helps to make early interventions in repair decisions, prevent further crack expansion and rail failure, and reduce repair costs. This method can also significantly reduce the workload and time cost of manual inspections, meeting the efficiency requirements of high-speed rail inspections.

[0049] In an optional example of this embodiment, the oblique cracks and block loss of the oblique peeling cracks and the rail surface block loss are quantitatively calculated to obtain the corresponding block loss depth value. The identification of the oblique crack damage level is achieved through deep learning and image recognition processing methods, such as Figure 3 As shown in the figure, the morphological characteristics and damage extent of different grades of oblique cracks are analyzed, and the damage grade is determined by combining the damage depth obtained through nondestructive testing. For quantified calculation of chipping damage, the area of ​​the chipping is calculated using 2D images. 3D point cloud data of the chipping is also located, and the chipping depth is calculated based on the point cloud contour data. Calculating the chipping area using 2D images and the chipping depth using 3D point cloud data allows for a more accurate assessment of the severity of oblique cracks and chipping. This provides a more intuitive reference for maintenance personnel through the use of image features and nondestructive testing characteristics, effectively reducing missed detections and misjudgments.

[0050] In an optional embodiment of the present invention, rail oblique peeling cracks are classified into different levels based on characteristics of different development stages, including:

[0051] When the morphological characteristics are linear cracks with an angle to the grinding marks or the driving direction, and the damage depth of the non-destructive test is less than 2mm, the corresponding rail surface cracks are classified as rail surface microcracks, such as Figure 6-1 As shown;

[0052] When the morphological characteristics are obvious single oblique cracks, and the light band does not widen, and the damage depth of non-destructive testing is less than or equal to 5mm, the corresponding rail oblique peeling crack is classified as oblique crack level I, such as Figure 6-2 As shown;

[0053] When the morphological characteristics are that the light band widens on one side and the crack has a V-shaped feature, and the damage depth of the non-destructive test is 2.0 to 8.0 mm, the corresponding rail oblique peeling crack is classified as oblique crack grade II. Figure 6-3 As shown;

[0054] When the morphological characteristics are that the light band widens on both sides and the crack has a V-shaped feature, and the damage depth of the non-destructive test is greater than 5.0 mm, the corresponding rail oblique peeling crack is classified as oblique crack rating level III, such as Figure 6-4 As shown;

[0055] When the morphological characteristics are that the light band widens on both sides, there is chipping in the crack center, and the depth of the chipping is less than 0.3mm, and the damage depth of the non-destructive test is 2.0-8.0mm, the corresponding rail oblique peeling crack is classified as oblique crack with chipping level I. Figure 6-5 As shown;

[0056] When the morphological feature is that the crack has chipping, and the chipping depth is 0.3-1.0mm, and the damage depth of the non-destructive test is greater than 5.0mm, the corresponding rail oblique peeling crack is classified as oblique crack and chipping rating is II. Figure 6-6 As shown;

[0057] When the morphological characteristics are continuous chipping of cracks, the chipping depth is greater than 1.0mm, and the damage depth of non-destructive testing is greater than 5.0mm, the corresponding rail oblique peeling crack is classified as oblique crack and chipping level III. Figure 6-7 shown.

[0058] Specifically, by using relevant track rapid comprehensive inspection instruments equipped with 2D and multi-channel eddy current flaw detectors or 2D+3D cameras and multi-channel eddy current flaw detectors, images of the rail surface status are collected, and damage depth data is collected at the same time. The oblique peeling cracks on the rails are identified and marked using machine vision-based data analysis software, and further combined with non-destructive testing data and rail surface image features, the oblique peeling cracks on the rails are graded based on the characteristics of different development stages. The rating method of the present invention integrates multi-dimensional data, and through deep learning and image recognition processing methods, it can more accurately identify the different damage states of the oblique peeling cracks on the rails, and quantitatively analyze the damage morphology characteristics and damage degree of oblique cracks of different grades to achieve accurate rating.

[0059] In an optional embodiment of the present invention, the rail surface oblique peeling crack rating method further includes:

[0060] Use data playback to supplement missed damage data through manual review;

[0061] The rail oblique peeling cracks were re-inspected on site using non-destructive testing.

[0062] Specifically, as shown in Figure 6, after the oblique crack identification is completed, data playback is manually reviewed. The process includes: image and eddy current data playback review and on-site verification:

[0063] 1. Data playback:

[0064] After the recognition is completed, the data is replayed and manually reviewed, such as Figure 5As shown in the figure, a comprehensive analysis is conducted on the characteristics of the oblique peeling crack and the range of the collected eddy current signal. In addition, the more serious oblique peeling crack has a low collapse depression at the damage site. The 3D image information of the rail can be combined to determine the occurrence of rail surface widening and provide a value for the block drop depth. The damage identified by machine vision can be reviewed, corrected, and supplemented according to the corresponding damage severity type.

[0065] When the oblique crack depth data is missing, refer to Figure 5 The damage maps shown are used to compare and rate the damage morphology characteristics of each damage level.

[0066] At this point, the judgment and marking of the damage degree of the oblique peeling cracks in the overall inspection data are completed.

[0067] 2. On-site review

[0068] After all the diagonal peeling cracks have been graded and identified, the railway skylight time is used to go to the section that needs special attention to find the corresponding damage based on the mileage information and damage morphology in the picture. Non-destructive testing methods are used on site, including ultrasonic thickness measurement, magnetic leakage and other non-destructive testing to conduct on-site review of each type of damage.

[0069] The rating method proposed in this invention significantly improves the accuracy and reliability of test results through manual review of data playback and on-site non-destructive testing re-inspection. The data playback function enables inspectors to conduct a second review of damage identified by machine vision, and, combined with eddy current signals and 2D image information, effectively supplement and correct possible missed detections and misjudgments, ensuring the accuracy of damage ratings. On-site non-destructive testing re-inspection further verifies the test results, especially in the measurement of damage depth and chip size, providing more intuitive and accurate data support. This dual verification mechanism not only reduces the missed detection rate and misjudgment rate, but also enhances the credibility of the test results, providing a solid data foundation for rail maintenance decisions, thereby ensuring the safety of train operation and reducing the overall cost of rail maintenance.

[0070] The detailed explanations of the above-mentioned embodiments are intended only to explain the present invention so as to facilitate a better understanding of the present invention. However, these descriptions cannot be interpreted as limiting the present invention for any reason. In particular, the various features described in different embodiments may also be arbitrarily combined with each other to form other embodiments. Unless otherwise clearly described, these features should be understood to be applicable to any embodiment and are not limited to the described embodiments.

Claims

1. A method for detecting and rating rail oblique cracks based on machine vision and non-destructive testing, characterized in that: Capture images of the rail surface and perform non-destructive testing on the rail damage depth data; Based on machine vision, the rail oblique peeling crack is identified in the acquired image and the morphological features of the oblique peeling crack are extracted; The rail oblique peeling cracks are classified into different levels based on the characteristics of different development stages by combining the morphological features and the damage depth data obtained by non-destructive testing.

2. The rail surface oblique peeling crack rating method according to claim 1, characterized in that: The acquired images are 2D images and / or 3D images.

3. The rail surface oblique peeling crack rating method according to claim 1, characterized in that: Damage depth data is collected through ultrasonic testing, magnetic flux leakage testing, radiographic testing and / or eddy current testing.

4. The rail surface oblique peeling crack rating method according to claim 3, characterized in that: The damage depth data collection covers the entire wheel-rail contact area on the transverse rail surface.

5. The rail surface oblique peeling crack rating method according to claim 1, characterized in that: The identification results of the oblique linear peeling cracks of the rail are oblique crack level I, oblique crack level II, crack level III, oblique crack and chipping level I, oblique crack and chipping level II and oblique crack and chipping level III respectively.

6. The rail surface oblique peeling crack rating method according to claim 5, characterized in that: The identified oblique peeling cracks of the rail are marked.

7. The rail surface oblique peeling crack rating method according to claim 5, characterized in that: The oblique cracks and chipping of the oblique peeling cracks and the chipping of the rail surface are quantitatively calculated to obtain corresponding chipping depth values.

8. The rail surface oblique peeling crack rating method according to claim 1, wherein: Rail oblique peeling cracks are classified into different levels based on the characteristics of different development stages, including: When the morphological feature is a linear crack that is at an angle to the grinding mark or the driving direction, and the damage depth detected by non-destructive testing is less than 2 mm, the corresponding oblique linear peeling crack of the rail is classified as a rail surface microcrack; When the morphological feature is a distinct single oblique crack, the light band does not widen, and the damage depth of the non-destructive test is less than or equal to 5 mm, the corresponding oblique linear peeling crack of the rail is classified as oblique crack level I; When the morphological feature is that the light band widens on one side and the crack has a V-shaped feature, and the damage depth of the non-destructive test is 2.0 to 8.0 mm, the corresponding rail oblique peeling crack is classified as oblique crack level II; When the morphological feature is that the light band widens on both sides and the crack has a V-shaped feature, and the damage depth of the non-destructive test is greater than 5.0 mm, the corresponding rail oblique peeling crack is classified as oblique crack rating level III; When the morphological characteristics are that the light band widens on both sides, a chip appears in the center of the crack, and the depth of the chip is less than 0.3 mm, and the damage depth of the non-destructive test is 2.0-8.0 mm, the corresponding rail oblique peeling crack is classified as oblique crack with chipping level I; When the morphological feature is that the crack has chipping, and the depth of the chipping is 0.3-1.0 mm, and the damage depth of the non-destructive test is greater than 5.0 mm, the corresponding rail oblique peeling crack is classified as an oblique crack and the chipping rating is graded as Level II; When the morphological feature is that the crack has continuous chipping, the chipping depth is greater than 1.0 mm, and the damage depth of the non-destructive testing is greater than 5.0 mm, the corresponding rail oblique peeling crack is classified as oblique crack and chipping level III.

9. The rail surface oblique peeling crack rating method according to claim 1, wherein: The rail surface oblique peeling crack rating method further includes: Use data playback to supplement missed damage data through manual review; The rail oblique peeling cracks are re-inspected on site using non-destructive testing.

Images