Wheel-rail adhesion coefficient measuring method and system based on image recognition

Through image recognition technology, the wheel and rail adhesion coefficient is measured in real time, which solves the problem of difficulty in accurately obtaining the wheel and rail adhesion coefficient in the prior art, and effectively controls wheel slip and wear, and improves the traction and braking performance of train operations.

CN119246403APending Publication Date: 2025-01-03SOUTHWEST JIAOTONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411072268.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2025-01-03

AI Technical Summary

Technical Problem

The prior art is difficult to accurately obtain the wheel and rail adhesion coefficient in a timely manner, especially in complex environments, such as rainy days or curved areas, resulting in increased wheel slippage and wear, and it is difficult to achieve effective traction and braking control.

Method used

Using the wheel and rail adhesion coefficient measurement method based on image recognition, the wheel rail contact position image is taken by installing a camera under the vehicle, and using feature point recognition and image processing technology to calculate the wheel speed and running speed in real time, combined with the sticky and slip characteristic curve in the database. Real-time measurement of the adhesion coefficient without contact.

Benefits of technology

It realizes accurate and real-time measurement of the wheelset adhesion coefficient, supports anti-air rotation and anti-slip control for train operation, improves operational efficiency and ride comfort, and reduces operating costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119246403A_ABST
    Figure CN119246403A_ABST
Patent Text Reader

Abstract

The invention discloses a wheel-rail adhesion coefficient determination method and system based on image recognition, and the method employs a CCD industrial camera to achieve the non-contact measurement of the wheel-rail operation speed, and employs an image processing technology to process the wheel-rail image information in real time to obtain the wheel-rail operation speed, thereby obtaining the wheel-rail creep rate. According to the wheel rail adhesion creep characteristic curve in the storage database, the adhesion coefficients of all the wheels and the steel rail contact area when the train runs at the current speed are obtained, the adhesion coefficient of any wheel set is obtained, and non-contact real-time measurement of the adhesion coefficients of the wheel sets is achieved; and data support is provided for improving an anti-slip or anti-slip control method and strategy for train operation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of rail transportation, and in particular relates to a wheel-rail adhesion coefficient measurement method and system based on image recognition. Background Art

[0002] Traction and braking of rail trains are highly related to wheel-rail adhesion. When wheel-rail adhesion is insufficient, it is easy to cause idling and slipping. The vehicle cannot effectively speed up under traction conditions, and cannot brake in a shorter distance under braking conditions, resulting in energy waste. At the same time, it will also cause scratches on the rails or wheel treads, and wear and tear, increasing operating costs. The resulting intensified vibration and impact between the wheels and rails will produce strong noise and reduce ride comfort. The accurate acquisition of the adhesion coefficient is of great significance. When the train vehicle is traveling on a section with a low adhesion coefficient, such as rainy days, flooded sections or curved areas, the friction between the wheels and the rails will decrease, resulting in a decrease in adhesion between the wheels and rails. This will cause the wheels to slip, increase the wear between the wheels and the rails, and also increase the friction between the wheels and rails, further reducing the adhesion, forming a vicious circle. The adhesion coefficient can be obtained through calculation and testing. The calculation needs to consider the influence of running speed, wheel-rail surface conditions and environmental factors, and it is difficult to restore the actual situation of the line; the test can only be tested through indoor test benches, and it is difficult to obtain the real-time adhesion coefficient distribution of the running line. Summary of the invention

[0003] In order to overcome the above-mentioned shortcomings, a wheel-rail adhesion coefficient measurement method and system based on image recognition are proposed. The system can be installed under the vehicle to capture the wheel-rail contact position image, and use image recognition technology to extract and identify feature points to obtain the wheel speed and running speed in real time. According to the stick-slip characteristic curve stored in the database, the adhesion coefficient of any wheelset is obtained, and the non-contact real-time measurement of the wheelset adhesion coefficient is realized, providing data support for improving the anti-spin or anti-skid control methods and strategies for train operation.

[0004] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is: to provide a method and system for measuring the wheel-rail adhesion coefficient based on image recognition. The steps include: 1) Obtaining wheel-rail image information: using a camera to continuously shoot the contact part of the train wheel and rail to obtain the wheel and track images; 2) Image processing: first filter out the interference noise from the wheel and track images obtained by shooting, 3) Feature point identification: differentiating two consecutive frames of images to obtain a differential image, and then performing grayscale transformation and edge extraction on the differential image to identify the feature points on the wheel and rail; 4) Solving the creep rate: according to the time difference between shooting two consecutive images, the moving distance of the feature points on the wheel and rail, the wheel speed is calculated respectively. and train speed Creep rate Calculation formula: Where R is the actual radius of the wheel; 5) Solve the adhesion coefficient: According to the wheel-rail adhesion creep characteristic curve stored in the database, the target wheel-rail real-time adhesion coefficient is obtained.

[0005] According to the wheel-rail adhesion coefficient determination method based on image recognition described in the present invention, a further preferred technical solution is: in step 1), an industrial camera with adjustable frame rate is used for shooting, and the number of frames shot when the wheel rotates one circle is more than 5 frames. The CCD image sensor converts the light intensity reflected by the wheel and the rail into an electrical signal, obtains digital information through the A / D conversion module, and then transmits the data to a remote computer through wireless transmission.

[0006] According to the method for determining the wheel-rail adhesion coefficient based on image recognition described in the present invention, a further preferred technical solution is: in step 3), the recognition rule of the characteristic points of the wheel part is to mark 36 recognition points on the wheel rim, one recognition point every 10 degrees, and the recognition point is the mark closest to the wheel-rail contact point in the forward direction; the recognition rule of the characteristic points of the undercarriage sleeper is based on the continuous distribution of the rail fasteners with a spacing of 60 cm, and the recognition point is the sharp corner on the inner rear side of the fastener closest to the wheel-rail contact point in the forward direction.

[0007] According to the wheel-rail adhesion coefficient determination method based on image recognition described in the present invention, its further preferred technical solution is: in step 2), the image processing step is specifically: setting the image, the following is the minimum setting standard, setting the pixel tolerance value of the algorithm to 7, the end judgment value to 5, and the pixel tolerance is used to eliminate non-zero invalid pixels. The specific steps are as follows: Step 1: Read the differential image matrix, set flagl = 0, flag2 = 0, scan the first column, and perform tolerance processing on each pixel. If a valid pixel is detected, execute step 3, otherwise execute step 2; Step 2: Scan the next column, set flagl = 0, flag2 = 0, and perform tolerance processing on each pixel. If a valid pixel is detected, execute step 3; if there is no valid pixel, repeat this step; Step 3: Record the position (x, y) of the valid pixel, and determine the value of the pixel below the point; Step 4: If it is a valid pixel, repeat step 3; Step 5: If it is an invalid pixel, record the number of times the invalid pixel appears and determine the value of the pixel below it; Step 6: If it is a valid pixel, repeat step 3; If it is an invalid pixel and the number of invalid pixel occurrences is less than 5, repeat step 5; Step 7: If the number of invalid pixel occurrences is greater than 5, calculate the average position (x, y) of the valid pixel and record the number of valid pixel columns; If the number of valid pixel columns is greater than 5, determine the change trend of the ordinate Y; If the change trend of the valid pixel changes, set flag2 = 1; If flagl = 0 and flag2 = 0, store (x, y) in the first array and set flagl = 1, otherwise store (xy) in the second array and repeat step 2 According to the wheel-rail adhesion coefficient determination method based on image recognition described in the present invention, a further preferred technical solution is: in step 5), in order to improve efficiency and accuracy, the wheel-rail adhesion creep characteristic curve is auxiliary selected according to the train running speed, axle weight, slope and wheel-rail contact state, and then the target wheel-rail real-time adhesion coefficient is calculated from the curve.

[0008] According to the method for measuring wheel-rail adhesion coefficient based on image recognition described in the present invention, a further preferred technical solution is that the wheel-rail contact state includes dry, wet, oily, and ice and snow.

[0009] According to the wheel-rail adhesion coefficient measurement method based on image recognition described in the present invention, a further preferred technical solution is: in step 1), any wheelset of the train is selected to obtain wheel-rail image information.

[0010] A wheel-rail adhesion coefficient measurement system based on image recognition, comprising: a wheel-rail image acquisition unit: selecting a suitable frame rate for shooting according to the vehicle speed displayed on the control console, and transmitting the image to a remote computer by wireless means; a feature extraction unit: extracting feature points of the wheel and the rail at the same position, and obtaining the wheel rotation displacement and the rail relative displacement of two adjacent frames of images through edge detection at different scales, and then obtaining the wheel rotation speed and the running speed according to the time difference between the two frames of images; a creep rate solving unit: used for solving the creep rate; a memory unit: used for storing a large number of stick-slip characteristic curves; an output and display unit: selecting a suitable stick-slip characteristic curve in the memory unit according to the creep rate to obtain the real-time adhesion coefficient.

[0011] According to the wheel-rail adhesion coefficient measurement system based on image recognition described in the present invention, its further preferred technical solution is: the wheel-rail image acquisition unit includes an industrial camera, an auxiliary lighting device, a fixing device and a data wireless transmission device; the fixing device is used to fix the industrial camera, the auxiliary lighting device and the data wireless transmission device near the wheel-rail, the industrial camera is used for shooting, the auxiliary lighting device is used for shooting fill light, and then the image data obtained by shooting is transmitted to a remote computer through the wireless transmission device.

[0012] According to the wheel-rail adhesion coefficient measurement system based on image recognition described in the present invention, its further preferred technical solution is: the frame rate of the industrial camera is adjustable, covering a range of 0-240f / s, 300,000-20 million pixels, and the gain, exposure time and white balance are programmable.

[0013] Compared with the prior art, the technical solution of the present invention has the following advantages / benefits: 1. The system can be installed under the vehicle to capture images of the wheel-rail contact position. By using image recognition technology, the wheel speed and running speed can be obtained in real time through the extraction and recognition of feature points. The adhesion coefficient of any wheelset can be obtained based on the stick-slip characteristic curve stored in the database, and the non-contact real-time measurement of the adhesion coefficient of the wheelset can be achieved, providing data support for improving anti-spin or anti-skid control methods and strategies for train operation.

[0014] 2. This measurement method is not limited to indoor test benches and can be applied to any running vehicle. Relying on image recognition technology and stick-slip curve database, it can accurately obtain the real-time adhesion coefficient of any wheelset with high accuracy and realize application in real scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a schematic diagram of characteristic identification points of a wheel-rail adhesion coefficient determination method and system based on image recognition of the present invention.

[0017] Figure 2 These are the stick-slip curves for different normal forces under dry conditions, straight segments, and constant speed conditions. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. Therefore, the detailed description of the embodiments of the present invention provided below is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention.

[0019] Embodiment 1: like Figure 1As shown, a method and system for determining the wheel-rail adhesion coefficient based on image recognition. The steps include: 1) Obtaining wheel-rail image information: using a camera to continuously shoot the wheel-rail contact area of ​​the train to obtain the wheel and track images; using an industrial camera with adjustable frame rate to shoot, the number of frames shot when the wheel rotates one circle is more than 5 frames, the CCD image sensor (camera) converts the light intensity reflected by the wheel and rail into an electrical signal, obtains digital information through the A / D conversion module, and then transmits the data to a remote computer through wireless transmission. At the same time, the present invention selects any wheelset of the train to obtain wheel-rail image information, that is, the industrial camera is set near any wheelset to be detected to shoot the wheel-rail contact area of ​​the wheelset; 2) Image processing: First, the interference noise is filtered out of the wheel and track images obtained by shooting. The specific steps of image processing are: setting the image, the following is the minimum setting standard, setting the pixel tolerance value of the algorithm to 7, and the end judgment value to 5. The pixel tolerance is used to eliminate non-zero invalid pixels. The specific steps are as follows: Step 1: Read the differential image matrix, set flagl = 0, flag2 =0, scan the first column, perform tolerance processing on each pixel, if a valid pixel is detected, execute step 3, if not, execute step 2; Step 2: Scan the next column, set flagl=0, flag2=0, perform tolerance processing on each pixel, if a valid pixel is detected, execute step 3; if there is no valid pixel, repeat this step; Step 3: Record the position (x, y) of the valid pixel and determine the value of the pixel below it; Step 4: If it is a valid pixel, repeat step 3; Step 5: If it is an invalid pixel, record the number of times the invalid pixel appears and determine the value of the pixel below it; Step 6: If it is a valid pixel, repeat step 3; If it is an invalid pixel and the number of invalid pixel occurrences is less than 5, repeat step 5; Step 7: If the number of invalid pixel occurrences is greater than 5, calculate the average position (x, y) of the valid pixel and record the number of valid pixel columns; If the number of valid pixel columns is greater than 5, determine the change trend of the ordinate Y; If the change trend of the valid pixel changes, set flag2 =1; if flagl=0 and flag2=0, store (x, y) in the first array and set flagl =1, otherwise store (xy) in the second array and repeat step 2; In fact, due to the interference and influence of various noises during the generation, transmission and transformation of images, the image quality will inevitably be reduced to varying degrees, which brings difficulties to image processing. Image preprocessing mainly processes the basic feature information in the original image to obtain a clear and accurate image for subsequent image processing. The image processing step is only for preprocessing the captured image, and other similar processing methods can be used. 3) Feature point recognition: The difference between two consecutive frames of images is obtained by differential, and then the differential image is gray-scale transformed and edge extracted to identify the feature points on the wheel and rail; the recognition rule of the feature points of the wheel part is to mark 36 recognition points on the wheel rim, one recognition point every 10 degrees, and the recognition point is the mark closest to the wheel-rail contact point in the forward direction; the recognition rule of the feature points of the undercarriage rail sleeper is based on the continuous distribution of the rail fasteners with a spacing of 60cm, and the recognition point is the sharp corner of the inner rear of the fastener closest to the wheel-rail contact point in the forward direction. Of course, the setting of the feature points only plays a recognition role, and other recognition points can also be set according to the wheel-rail structure. 4) Calculate the creep rate: The wheel speed is calculated based on the time difference between two consecutive images and the moving distance of the feature points on the wheel and rail. and train speed Creep rate Calculation formula: Where R is the actual wheel radius; 5) Solve the adhesion coefficient: According to the wheel-rail adhesion creep characteristic curve stored in the database, the target wheel-rail real-time adhesion coefficient is obtained. In order to improve efficiency and accuracy, the wheel-rail adhesion creep characteristic curve is selected according to the train running speed, axle weight, slope and wheel-rail contact state (contact environment), and then the target wheel-rail real-time adhesion coefficient is calculated from the curve. The wheel-rail contact state includes dry, wet, oily, ice and snow, and can also be based on actual conditions.

[0020] A wheel-rail adhesion coefficient measurement system based on image recognition, comprising: a wheel-rail image acquisition unit: selecting a suitable frame rate for shooting according to the vehicle speed displayed on the control console, and transmitting the image to a remote computer by wireless means; Feature extraction unit: extracts feature points at the same position of the wheel and the rail, obtains the wheel rotation displacement and the relative displacement of the rail in two adjacent frames of images through edge detection at different scales, and then obtains the wheel speed and running speed based on the time difference between the two frames of images; Creep rate solving unit: used to solve the creep rate; Memory unit: used to store a large number of stick-slip characteristic curves; Output and display unit: selects the appropriate stick-slip characteristic curve in the memory unit according to the creep rate to obtain the real-time adhesion coefficient.

[0021] The wheel-rail image acquisition unit includes an industrial camera, an auxiliary lighting device, a fixing device and a data wireless transmission device; the fixing device is used to fix the industrial camera, the auxiliary lighting device and the data wireless transmission device near the wheel-rail, the industrial camera is used for shooting, the auxiliary lighting device is used for shooting fill light, and then the image data obtained by shooting is transmitted to a remote computer through the wireless transmission device.

[0022] The industrial camera has an adjustable frame rate, covering a range of 0-240f / s, 300,000-20 million pixels, and programmable control of gain, exposure time and white balance. The industrial camera can also be replaced with other shooting equipment as long as the shooting standards are met.

[0023] Example 2: Taking the actual situation as the scenario, the train runs on the line at 300km / h, and the adhesion coefficient of a wheelset needs to be measured. The image acquisition unit is arranged at a wheelset, and the wheel-rail contact image can be captured. The wheel is a new wheel with a diameter of 920mm and a wheel circumference of 2.89m. The maximum number of circles the wheel rotates in 1s at a speed of 300km / h is 28.84 circles. Considering the system accuracy, at least 5 or 6 frames of pictures should be taken for one circle of wheel rotation, and the frame rate of the industrial camera is required to be adjusted to a minimum of 174f / s. After the image is captured, it is sent to the remote computer via wireless transmission.

[0024] After the remote computer receives the image, it enters the feature extraction unit, and after preprocessing, the image recognizability is enhanced. After feature point extraction and precise positioning technology of large and small scale edge points, the wheel rotation displacement and relative displacement of the rail are obtained.

[0025] Subsequently, the creep rate solving unit solves the wheel rotation speed and running speed according to the feature point recognition results of the feature extraction unit, namely the wheel rotation displacement and the relative displacement of the rail, according to the time difference of image shooting, so as to calculate the wheel-rail creep rate.

[0026] The memory unit uses artificial intelligence technology to comprehensively consider the operating speed, axle weight, slope, and contact state (dry, wet, oily, ice and snow), assist in selecting the appropriate stick-slip characteristic curve, and obtain the adhesion coefficient based on the creep rate.

[0027] The output and display unit outputs the real-time adhesion coefficient of the target wheel-rail, which is arranged by time or mileage. The adhesion coefficient of the wheelset for the entire line can be obtained, which is used to improve the anti-spin or anti-skid control methods and strategies.

[0028] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0029] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0030] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature.

[0031] The above are only preferred embodiments of the present invention. It should be noted that the above preferred embodiments should not be regarded as limiting the present invention, and the protection scope of the present invention should be based on the scope defined by the claims. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the spirit and scope of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A wheel-rail adhesion coefficient measurement method based on image recognition, characterized in that: The steps include: 1) Obtaining wheel-rail image information: Using a camera to continuously shoot the wheel-rail contact area of ​​the train to obtain wheel and rail images; 2) Image processing: First, filtering out interference noise from the wheel and rail images obtained by shooting; 3) Feature point identification: Differentiating two consecutive frames of images to obtain a differential image, then performing grayscale transformation and edge extraction on the differential image to identify the feature points on the wheel and rail; 4) Solving the creep rate: According to the time difference between shooting two consecutive images and the moving distance of the feature points on the wheel and rail, the wheel speed is calculated respectively. and train speed Creep rate Calculation formula: Where R is the actual wheel radius; 5) Solve the adhesion coefficient: According to the wheel-rail adhesion creep characteristic curve stored in the database, the target wheel-rail real-time adhesion coefficient is obtained.

2. The wheel-rail adhesion coefficient measurement method based on image recognition according to claim 1, characterized in that: In step 1), an industrial camera with adjustable frame rate is used for shooting. The number of frames shot for one wheel rotation is more than 5. The CCD image sensor converts the light intensity reflected by the wheel and rail into an electrical signal, obtains digital information through the A / D conversion module, and then transmits the data to a remote computer through wireless transmission.

3. The wheel-rail adhesion coefficient measurement method based on image recognition according to claim 1, characterized in that: In step 3), the recognition rule of the feature points of the wheel is to mark 36 recognition points on the wheel rim, one recognition point every 10 degrees, and the recognition point is the mark closest to the wheel-rail contact point in the forward direction; the recognition rule of the feature points of the undercarriage sleeper is based on the continuous distribution of the rail fasteners with a spacing of 60 cm, and the recognition point is the sharp corner on the inner rear side of the fastener closest to the wheel-rail contact point in the forward direction.

4. The wheel-rail adhesion coefficient measurement method based on image recognition according to claim 1, characterized in that: In step 2), the image processing steps are specifically as follows: set the image, the following is the minimum setting standard, set the algorithm's pixel tolerance value to 7, the end judgment value to 5, and the pixel tolerance is used to eliminate non-zero invalid pixels. The specific steps are as follows: Step 1: Read the differential image matrix, set flagl = 0, flag2 = 0, scan the first column, and perform tolerance processing on each pixel. If a valid pixel is detected, execute step 3, otherwise execute step 2; Step 2: Scan the next column, set flagl = 0, flag2 = 0, and perform tolerance processing on each pixel. If a valid pixel is detected, execute step 3; if there is no valid pixel, repeat this step; Step 3: Record the position (x, y) of the valid pixel and determine the value of the pixel below the point; Step 4: If it is a valid pixel, repeat step 3; Step 5: If it is an invalid pixel, record the number of times the invalid pixel appears and determine the value of the pixel below the point; Step 6: If it is a valid pixel, repeat step 3; If it is an invalid pixel and the number of times the invalid pixel appears is less than 5, repeat step 5; Step 7: If the number of invalid pixels is greater than 5, calculate the average position (x, y) of the valid pixels and record the number of valid pixel columns; if the number of valid pixel columns is greater than 5, determine the change trend of the vertical coordinate Y; if the change trend of the valid pixels changes, set flag2 = 1; if flagl = 0 and flag2 = 0, store (x, y) in the first array and set flagl = 1, otherwise store (xy) in the second array and repeat step 2.

5. The wheel-rail adhesion coefficient measurement method based on image recognition according to claim 1, characterized in that: In step 5), in order to improve efficiency and accuracy, the wheel-rail adhesion creep characteristic curve is selected according to the train running speed, axle weight, slope and wheel-rail contact state, and then the target wheel-rail real-time adhesion coefficient is calculated from the curve.

6. The wheel-rail adhesion coefficient measurement method based on image recognition according to claim 5, characterized in that: Wheel-rail contact conditions include dry, wet, oily, and ice and snow.

7. The wheel-rail adhesion coefficient measurement method based on image recognition according to claim 1, characterized in that: In step 1), select any wheelset of the train to obtain wheel-rail image information.

8. A wheel-rail adhesion coefficient measurement system based on image recognition, characterized in that: It includes: Wheel-rail image acquisition unit: selects the appropriate frame rate for shooting according to the vehicle speed displayed on the control console, and transmits the image to the remote computer wirelessly; Feature extraction unit: extracts the feature points of the wheel and rail at the same position, obtains the wheel rotation displacement and rail relative displacement of two adjacent frames of images through edge detection at different scales, and then obtains the wheel speed and running speed based on the time difference between the two frames of images; Creep rate solving unit: used to solve the creep rate; Memory unit: used to store a large number of stick-slip characteristic curves; Output and display unit: According to the creep rate, a suitable stick-slip characteristic curve in the memory unit is selected to obtain the real-time sticking coefficient.

9. The wheel-rail adhesion coefficient measurement system based on image recognition according to claim 8, characterized in that: The wheel-rail image acquisition unit includes an industrial camera, an auxiliary lighting device, a fixing device and a data wireless transmission device; the fixing device is used to fix the industrial camera, the auxiliary lighting device and the data wireless transmission device near the wheel-rail, the industrial camera is used for shooting, the auxiliary lighting device is used for shooting fill light, and then the image data obtained by shooting is transmitted to a remote computer through the wireless transmission device.

10. The wheel-rail adhesion coefficient measurement system based on image recognition according to claim 9, characterized in that: The industrial camera has an adjustable frame rate, covering a range of 0-240f / s, 300,000-20 million pixels, and programmable control of gain, exposure time and white balance.