Automatic detection device and method for rack surface wear of rack rail

Abstract

This invention relates to the field of rail transit inspection technology, and discloses an automated detection device and method for tooth surface wear of a gear rail. The automated detection device for tooth surface wear includes a gear rail structure, rails, and a trolley frame. It further includes: a traveling unit mounted on the trolley frame for driving the detection device along the railway rails and for braking and maintaining a stationary position on slopes; and a tooth profile scanning device mounted on the trolley frame, which moves laterally, vertically, and rotates along the trolley frame to perform multi-distance, multi-position, and multi-angle scanning of the gear rail structure to obtain the three-dimensional tooth profile and tooth surface wear distribution. This invention can grasp the tooth surface wear state pattern that changes with different operating mileages, providing a focus for the operation and maintenance of gear rail railways. Furthermore, it can achieve non-contact, continuous, and automated scanning, improving detection efficiency and ensuring measurement accuracy.

Description

Technical Field

[0001] This invention relates to the field of rail transit inspection technology, specifically to an automated inspection device and method for tooth surface wear of rail gears. Background Technology

[0002] The rack and pinion railway system uses gears and racks for traction drive, which has a stronger climbing ability and a smaller track gauge than the conventional wheel-rail adhesion system. It is particularly suitable for mountainous environments with large gradients. A pair of gears is added at the transverse center of the two rails to mesh with the rack and pinion. When climbing steep gradients, the gear and rack meshing provides traction drive force.

[0003] Rack and pinion railways differ significantly from traditional wheel-rail systems due to their steeper track gradients and rack-and-pinion meshing. Furthermore, their open gear-and-pinion meshing, exposed to the elements, differs considerably from the sealed precision gears found in gearboxes. This makes it more difficult to guarantee effective grease lubrication, potentially leading to greater wear. Once tooth wear exceeds permissible limits, it directly alters the gear tooth profile, reducing meshing accuracy and transmission smoothness. This results in noticeable vibration and noise, decreased ride comfort, and even safety issues. Therefore, regular monitoring and prevention of tooth wear are crucial.

[0004] Current gear tooth inspection devices can detect surface defects in gear teeth, but they cannot record or reproduce the tooth profile. Consequently, they cannot compare and analyze the actual worn tooth profile during operation with the initial tooth profile, calculate the wear depth and width distribution of the tooth surface, and thus cannot grasp the characteristics of the wear depth and width distribution of the gear tooth surface, or the evolution of the gear tooth profile with the length of operation. In other words, existing technology can achieve tooth surface flaw detection, but it cannot meet the requirements of gear tooth profile scanning and accurate detection of tooth surface wear. In addition, it is difficult to achieve non-contact, continuous, and automated scanning of gear tooth wear, which not only affects the detection efficiency but also leads to large measurement deviations. Summary of the Invention

[0005] This invention provides an automated detection device and method for tooth surface wear of gear rails. The detection device scans and reproduces the worn tooth surface, compares it with the initial tooth profile, and calculates the wear depth and width distribution. This allows for understanding the wear pattern of tooth surfaces under different operating mileages, providing a focus for the operation and maintenance of gear rail railways. It also addresses the shortcomings of traditional tooth surface wear calculation methods. Furthermore, it enables non-contact, continuous, and automated scanning, improving detection efficiency while ensuring measurement accuracy. This solves the problems mentioned in the background art, such as the difficulty in understanding the wear depth and width distribution characteristics of gear rail tooth surfaces, the difficulty in understanding the evolution of gear rail tooth profiles with operating mileage, low detection efficiency, and poor measurement accuracy.

[0006] This invention provides the following technical solution: An automated detection device for tooth surface wear of a toothed rail includes a toothed rail structure, a rail, and a trolley frame. It further includes: a traveling unit mounted on the trolley frame for driving the detection device along the railway rail and for braking and maintaining a stationary position on slopes; a tooth profile scanning device mounted on the trolley frame, which moves laterally, vertically, and rotates along the trolley frame to perform multi-distance, multi-position, and multi-angle scanning of the toothed rail structure to obtain the three-dimensional tooth profile and tooth surface wear distribution; and an adhesive magnet device symmetrically mounted on the trolley frame for detecting the rail slope and adjusting the normal pressure between the traveling unit and the rail according to the slope.

[0007] As a preferred embodiment of the present invention, the tooth profile scanning device includes: a transverse linear magnetic track, which is installed parallel to the trolley frame, and a transverse linear actuator is slidably connected to the transverse linear magnetic track; an upper and lower linear magnetic track, which is installed on the transverse linear actuator, and an upper and lower linear actuator is installed on the upper and lower linear magnetic track; a bracket, which is installed on the upper and lower linear actuator; a scanning angle control cylinder, which includes a cylinder body and a first telescopic rod, the cylinder body being installed on the bracket, and the first telescopic rod being installed on the cylinder body; and a laser scanner, which includes a scanner body, a scanning window, a first rotary joint, and a second rotary joint, the scanner body being rotatably connected to the cylinder body and the first telescopic rod respectively through the first rotary joint and the second rotary joint, and the scanning window being opened on the scanner body.

[0008] An automated method for detecting wear on the tooth surface of a gear track includes the following steps: Step 1: Laser scanning data acquisition: The three-dimensional coordinates and wear distribution of the structure are obtained by scanning with a detection device; Step 2: Data preprocessing and noise reduction: Random noise is removed from the scanned noise and outlier data using a filtering algorithm, followed by outlier removal. Step 3: Alignment of tooth root fillet features: Solve for the optimal rigid body transformation within the fillet feature area to achieve the best match between the initial profile and the measured profile; Step 4: Normal wear calculation: After alignment, calculate the normal wear depth at each point; Step 5: Wear distribution analysis: Perform parametric analysis on the wear data, parameterize the tooth surface, and convert the tooth surface coordinates into dimensionless parameters; Step Six: Results Output and Visualization: The final output consists of quantified parameters and visualized results.

[0009] As a preferred embodiment of the present invention, in step one, the initial state of the structural contour point cloud is denoted as... The structural contour point cloud under the measured condition is denoted as

[0010] As a preferred technical solution of the present invention, in step two, noise reduction: , in, , These are the filtering parameters for the spatial and intensity domains, respectively; Outliers: , Where k is the neighborhood size and threshold is the distance threshold.

[0011] As a preferred embodiment of the present invention, in step three, the rounded corner feature region is identified based on the radius of curvature constraint: ,in, This represents the local radius of curvature of a point in the point cloud, in mm. Equivalent curvature expression: Where k represents the local curvature, defined as the reciprocal of the radius of curvature. ; Matching the initial profile with the measured profile: , Where R: 3×3 rotation matrix, describing the rotational transformation of the coordinate system; T: 3×1 translation vector, describing the translation transformation of the coordinate system; Initial outline point cloud China belongs to The k-th point in the region; : Measured profile point cloud Zhongyu The corresponding nearest point; Euclidean norm, representing the Euclidean distance between two points; Constraints: Ensure the orthogonality of the rotation matrices. The constraints ensure that R is a valid rotation matrix, removing scaling or mirror transformations, where... Let R be the transpose of the rotation matrix. I is a 3×3 identity matrix. det(R): Determinant of the rotation matrix R.

[0012] As a preferred embodiment of the present invention, in step three, the optimal rigid body transformation is solved by: obtaining the translation vector of the rotation matrix through SVD decomposition. Point cloud centralization: First, subtract the centroid of each point set in the R16 region from the point sets of the two point clouds. , ; Constructing the covariance matrix: Calculate the covariance matrix between two centered point clouds: H: 3×3 covariance matrix, which contains the spatial relationship between the two point clouds; SVD Decomposition (Singular Value Decomposition): Singular value decomposition of the bicovariance matrix H: ,in, U: A 3×3 left singular vector matrix, an orthogonal matrix. Σ: A 3×3 diagonal matrix containing singular values. V: A 3×3 right singular vector matrix, an orthogonal matrix. : Transpose of matrix V; Calculate the optimal rotation matrix: Calculate the optimal rotation matrix using U and V: ; Calculate the translation vector: After obtaining the rotation matrix R, the translation vector T is calculated using the centroid relation: Rotation matrix: Translation vector: ; Aligned point cloud: The aligned point cloud is obtained by performing a rigid body transformation on the measured point cloud. .

[0013] In a preferred embodiment of the present invention, in step four, after alignment is completed, the normal wear depth of each point is calculated. The normal vector is determined by the pressure angle α, which is a constant value. ; For each initial profile point In the aligned measured point cloud Find its nearest point The normal wear depth is: .

[0014] In a preferred embodiment of the present invention, in step five, the wear data is subjected to parameter analysis, the tooth surface is parameterized, and the tooth surface coordinates are converted into dimensionless parameters. Tooth height direction: Tooth width direction ,in, u: Dimensionless tooth height parameter, range is , v: Dimensionless tooth width parameter, range is , : The y-coordinate of the tooth root position (lowest point). : The y-coordinate of the tooth tip (highest point). : The z-coordinate of the starting position of the tooth width : The z-coordinate of the end position of the tooth width; Establish the relationship between wear depth and location parameters: Wear distribution function: ; Average wear: ; Maximum wear: ; Standard deviation: ; Distributed along the tooth height direction: ,in, , for The number of points in the set, Δy: the width of the analysis window in the tooth height direction; Tooth width distribution: ,in, , for The number of points in the set, Δz: the width of the analysis window in the tooth width direction.

[0015] As a preferred embodiment of the present invention, in step six, the final output quantization parameters are: Overall average wear: ; Maximum wear location: ; Wear uniformity index: , range ; Visual output: 3D wear thermogram: Color mapping is a color gradient; Cross-sectional distribution curve: The changing trend.

[0016] Compared with the prior art, the present invention provides an automated detection device and method for tooth surface wear of gear tracks, which has the following beneficial effects: 1. In this automated detection equipment and method for tooth surface wear of the gear rail, the laser scanner can be fixed on a high-precision moving platform through the detection equipment to realize non-contact, continuous and automated scanning of tooth surface wear of the gear rail. No manual hand-held operation is required, which not only greatly reduces the labor intensity, but also effectively avoids measurement deviations caused by hand shaking, inaccurate positioning and other reasons.

[0017] 2. In this automated detection equipment and method for tooth surface wear of the gear rail, the time for a single detection is shortened through automated motion control and precise positioning. At the same time, the repeatability and accuracy of the scanned data are ensured, which is conducive to long-term tracking and comparative analysis of tooth surface wear. Moreover, the operation is simpler and the reliability is higher, making it suitable for continuous on-site operation and working conditions with varying environmental conditions.

[0018] 3. In this automated detection equipment and method for tooth surface wear of the rack rail, by comparing and analyzing the actual worn tooth profile during operation with the initial tooth profile, the wear depth and width distribution of the tooth surface can be calculated. This helps to understand the characteristics of the wear depth and width distribution of the rack rail tooth surface, as well as the evolution of the rack rail tooth profile with the number of kilometers of operation, thus providing a focus for the operation and maintenance of rack rail railways.

[0019] The parts of this device not covered herein are the same as or can be implemented using existing technologies. This invention can grasp the wear pattern of tooth surfaces that varies with different operating mileages, providing a focus for the operation and maintenance of gear rail railways. Moreover, it can achieve non-contact, continuous, and automated scanning, improving detection efficiency and ensuring measurement accuracy. Attached Figure Description

[0020] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, the elements or parts are not necessarily drawn to actual scale.

[0021] Figure 1 This is a three-dimensional schematic diagram of the present invention; Figure 2 This is a three-dimensional schematic diagram of the wheelset structure in this invention; Figure 3 This is a three-dimensional schematic diagram of the vehicle frame in this invention; Figure 4 This is a three-dimensional schematic diagram of the driving device in this invention; Figure 5 This is a three-dimensional schematic diagram of the tooth profile scanning device in this invention; Figure 6 This is a partial three-dimensional schematic diagram of the tooth profile scanning device in this invention; Figure 7 This is a three-dimensional schematic diagram of the scanner body in this invention; Figure 8 This is a three-dimensional schematic diagram of the axle box spring device in this invention; Figure 9 This is a three-dimensional schematic diagram of the adhesive magnet device in this invention; Figure 10 This is a flowchart illustrating the technical process of the present invention.

[0022] In the diagram: 100, wheelset structure; 110, wheel; 120, axle; 200, rack and pinion structure; 300, rail; 400, trolley frame; 410, side frame; 420, cross frame; 430, crossbeam; 500, drive unit; 510, belt; 520, motor; 530, sleeve; 540, output shaft; 600, tooth profile scanning device; 610, transverse linear magnetic track; 620, transverse linear mover; 630, upper and lower linear movers; 640, support frame; 650. Vertical linear magnetic track; 660. Scanning angle control cylinder; 661. Cylinder body; 662. First telescopic rod; 670. Laser scanner; 671. Scanner body; 672. Scanning window; 673. First rotary joint; 674. Second rotary joint; 700. Axle box spring device; 710. Rubber stack; 720. Axle box device; 800. Adhesive-enhancing magnet device; 810. Tilt sensor; 820. Second telescopic rod; 830. Permanent magnet. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1: Reference Figures 1-9 An automated detection device for tooth surface wear of a toothed rail includes a toothed rail structure 200, a steel rail 300, and a trolley frame 400. The trolley frame 400 includes side frames 410, cross frames 420, and cross beams 430. The cross frames 420 and cross beams 430 are installed between two side frames 410, and the cross beams 430 are located between two cross frames 420.

[0025] The trolley frame 400 forms a stable mobile platform, providing an installation reference for each functional module. The crossbeam 430 and the cross frame 420 form an I-shaped structure. The centrally positioned cross beam 430 enhances the overall torsional rigidity. The load is transferred to the wheels 110 through the side frame 410. The cross frame 420 and the cross beam 430 optimize the force flow path, making the integration of the drive unit 500 and the tooth profile scanning device 600 more compact and avoiding the vibration amplification effect caused by the cantilever structure. The modular frame design facilitates on-site disassembly and maintenance. The symmetrical structure ensures the stability of the center of gravity. The centrally positioned cross beam 430 provides the tooth profile scanning device 600 with an unobstructed working space across the tooth rail structure 200.

[0026] Reference Figure 2 , Figure 4 and Figure 8The traveling unit, mounted on the trolley frame 400, is used to drive the detection equipment along the railway rail 300 and to achieve braking and stationary holding on slopes. The traveling unit includes an axle box spring device 700, a wheelset structure 100, and a drive device 500. The axle box spring device 700 includes a rubber stack 710 and an axle box device 720. The rubber stack 710 is symmetrically mounted on the side frame 410, and the axle box device 720 is mounted on the rubber stack 710.

[0027] The axle box spring device 700 serves as a suspension buffer layer, the wheelset structure 100 is the walking execution unit, and the drive unit 500 is the power source—a three-level integrated architecture. The axle box spring device 700 filters high-frequency vibrations from the rail 300, the wheelset structure 100 provides guidance and support through contact between the wheels 110 and the rail 300, and the drive unit 500 converts rotational power into horizontal traction. These three components work together to achieve smooth movement, enabling the equipment to move autonomously on the track, adapt to slopes, and safely stop. The axle box spring device 700 protects the tooth profile scanning device 600 from vibration interference, and the wheelset structure 100 guides and ensures precise alignment between the scanning path and the toothed rail structure 200, while also facilitating maneuverability. For stability and measurement, the rubber stack 710 is an elastomer formed by vulcanized rubber, and the axle box device 720 is a housing structure that encloses the bearing. The two are stacked and installed. The rubber stack 710 absorbs vibration energy by utilizing the compression / shear deformation of rubber. Its nonlinear stiffness characteristics provide low stiffness to ensure comfort under no-load conditions, while the stiffness increases under heavy loads to limit vertical displacement. The axle box device 720 is connected to the axle 120 through bearings to transfer the vertical load to the rubber stack 710. The internal damping of the rubber stack 710 effectively attenuates high-frequency vibrations and prevents the laser scanner 670 from point cloud distortion caused by vibration. The symmetrical arrangement ensures uniform force on both sides and prevents the trolley frame 400 from being tilted, which would affect the scanning accuracy.

[0028] Reference Figure 2 The wheelset structure 100 includes: wheels 110 and axle 120. The axle 120 is rotatably connected to the axle box device 720. The wheels 110 are symmetrically mounted on the axle 120, and the distance between the two wheels 110 is the same as the spacing of the rails 300.

[0029] The axle 120 is a through-type solid round axle. The wheels 110 are fixed to both ends of the axle 120 by interference fit or bolts. The spacing of the wheels 110 strictly matches the spacing of the rails 300. The profile of the wheels 110 is selected to match the profile of the rails 300. The axle 120 rotates in the axle box device 720 through bearings, converting the driving torque into the traction force between the wheels and rails. The wheel track matching design eliminates lateral backlash, ensuring that the lateral relative position between the tooth profile scanning device 600 and the tooth rail structure 200 is constant. The centering accuracy directly determines the spatial reference accuracy of the scanning data, avoiding pseudo-wear errors caused by lateral offset of the tooth profile point cloud. The wheelset structure 100 is also the gravity bearing unit of the entire equipment, ensuring stable operation.

[0030] Reference Figure 4 The drive unit 500 includes: a motor 520 and a sleeve 530, the motor 520 being mounted on the crossbeam 420 via the sleeve 530; an output shaft 540 being mounted on the output end of the motor 520; and a belt 510 being sleeved between the output shaft 540 and the axle 120. The motor 520 is rigidly connected to the crossbeam 420 via the sleeve 530 flange. The output shaft 540 is arranged parallel to the axle 120. The belt 510 forms a flexible transmission closed loop. The output torque of the motor 520 is transmitted to the belt 510 via the output shaft 540. The belt 510 drives the axle 120 to rotate synchronously by friction. The sleeve 530 structure ensures that the axis of the motor 520 and the axis of the axle 120 have a fixed center distance, maintaining a constant tension of the belt 510. The scanning speed can be precisely adjusted by controlling the speed of the motor 520. The belt 510 transmission buffers the torque pulsation of the motor 520, avoiding uneven point cloud density caused by speed fluctuations during scanning. The flexible connection allows for a certain degree of installation error, reducing the requirements for processing accuracy.

[0031] Reference Figures 5-7 A tooth profile scanning device 600 is mounted on the trolley frame 400. The tooth profile scanning device 600 moves laterally and vertically and rotates along the trolley frame 400 to perform multi-distance, multi-position, and multi-angle scanning on the tooth rail structure 200 to obtain the three-dimensional tooth profile and tooth surface wear distribution of the tooth surface of the tooth rail structure 200. The tooth profile scanning device 600 includes: a transverse linear magnetic rail 610, which is mounted parallel to the trolley frame 400, and a transverse linear mover 620 is slidably connected to the transverse linear magnetic rail 610; an upper and lower linear magnetic rail 650, which is mounted on the transverse linear mover 620, and an upper and lower linear mover 630 is mounted on the upper and lower linear magnetic rail 650; and a bracket 640, which is mounted on the upper and lower linear mover 630.

[0032] The horizontal linear magnetic track 610 provides precise horizontal linear motion, while the vertical linear magnetic track 650 provides precise vertical linear motion. The horizontal linear magnetic track 610 and the vertical linear magnetic track 650 are linear motor stators with built-in magnets, and the horizontal linear actuator 620 and the vertical linear actuator 630 are sliders with windings, thereby driving the laser scanner 670 to move horizontally and vertically. The movement is smooth and vibration-free, suitable for high-precision scanning. The high-speed response characteristics support cross-tooth movement, improve detection efficiency, and provide a hardware foundation for complex scanning trajectories.

[0033] The scanning angle control cylinder 660 includes a cylinder body 661 and a first telescopic rod 662. The cylinder body 661 is mounted on a bracket 640, and the first telescopic rod 662 is mounted on the cylinder body 661.

[0034] The cylinder body 661 is a pressure chamber, and the first telescopic rod 662 is the piston rod output end, forming a linear actuator. Compressed air enters the cylinder body 661 chamber to push the piston, causing the first telescopic rod 662 to extend or retract linearly. The extension and retraction positions are infinitely adjustable by controlling the air pressure through a proportional valve, thereby driving the laser scanner 670 to rotate. The pneumatic drive has a fast response speed and a lightweight structure, making it suitable for frequent angle adjustments. The air pressure can be flexibly buffered to avoid damage to the laser scanner 670 from rigid collisions. Active angle control ensures that the laser beam is always perpendicularly incident in areas with varying curvature, such as tooth root fillets, improving point cloud quality.

[0035] The laser scanner 670 includes a scanner body 671, a scanning window 672, a first rotary joint 673 and a second rotary joint 674. The scanner body 671 is rotatably connected to the cylinder body 661 and the first telescopic rod 662 through the first rotary joint 673 and the second rotary joint 674, respectively. The scanning window 672 is opened on the scanner body 671.

[0036] The scanner body 671 is a closed shell, housing a laser, camera, and computing unit. The scanning window 672 is made of optical glass. The first rotary joint 673 and the second rotary joint 674 form two hinge points using a ball chain. When the first telescopic rod 662 extends or retracts, the laser scanner 670 swings around the first rotary joint 673 to achieve angle adjustment. The scanning window 672 faces the toothed track structure 200, and the laser line is projected from the scanning window 672 onto the tooth surface. The camera collects the deformed light stripe to calculate depth information, ensuring that the trajectory of angle adjustment is repeatable. The scanning angle has a non-linear relationship with the displacement of the first telescopic rod 662, which can be precisely controlled through calibration. The glass of the scanning window 672 is coated with an anti-reflective film to reduce laser energy loss and adapt to strong outdoor light environments.

[0037] Reference Figure 9 The adhesive magnet device 800 is symmetrically installed on the trolley frame 400 to detect the slope of the rail 300 and adjust the normal pressure between the traveling part and the rail 300 according to the slope. The adhesive magnet device 800 includes an angle sensor 810, a second telescopic rod 820 and a permanent magnet 830. The angle sensor 810 is symmetrically installed on the side frame 410. The second telescopic rod 820 is installed on the side of the side frame 410 near the angle sensor 810. The permanent magnet 830 is installed at the bottom end of the second telescopic rod 820 and is located above the rail 300.

[0038] The adhesion-enhancing magnet device 800 dynamically compensates for adhesion loss caused by changes in the slope of the rail 300, preventing slippage or runaway on the slope. It is symmetrically arranged on both sides of the frame 410. The tilt sensor 810 is an accelerometer type, and the second telescopic rod 820 is an electric push rod or hydraulic cylinder. The permanent magnet 830 is a strong magnetic block with a stainless steel sheath. The tilt sensor 810 measures the longitudinal slope angle of the rail 300 in real time. The control system calculates the required increase in adhesion force based on the slope and drives the second telescopic rod 820 to push the permanent magnet 830 down to a specific height from the rail surface. The attractive force generated by the permanent magnet 830 increases the wheel-rail normal pressure through the reaction of the trolley frame 400, thereby improving the adhesion coefficient, ensuring that the drive does not slip and the braking does not runaway, and ensuring uniform and stable scanning process. The magnetic attraction is non-contact, wear-free, and has a long lifespan. It can still provide basic attraction force in case of failure, increasing safety, while avoiding affecting the measurement field of view of the laser scanner 670 on the toothed rail structure 200.

[0039] Example 2: Reference Figure 10 Similar to Implementation 1, this paper proposes an automated detection method for tooth surface wear of gear tracks, comprising the following steps: Step 1: Laser scanning data acquisition: The three-dimensional coordinates and wear distribution of the structure are obtained by scanning with a detection device; Establish benchmark data and test data for wear assessment, place two scans in the same equipment coordinate system, quantify wear by comparing geometric deviations, eliminate systematic errors, focus only on service-related changes, and support long-term tracking and life prediction. In step one, the initial state of the structural contour point cloud is denoted as... The structural contour point cloud under the measured condition is denoted as .

[0040] Step 2: Data preprocessing and noise reduction: Random noise is removed from the scanned noise and outlier data using a filtering algorithm, followed by outlier removal. A bilateral filter is applied to smooth the data in both the spatial and intensity domains, preserving feature edges. The neighborhood distance of each point is calculated, and outliers far from the main point set are removed. The denoised point cloud retains subtle features such as tooth root fillets, avoiding excessive smoothing that could lead to false wear. Outlier removal prevents erroneous points from participating in the registration calculation, improving alignment accuracy. In step two, denoising is performed: , in, , These are the filtering parameters for the spatial and intensity domains, respectively; Outliers: , Where k is the neighborhood size and threshold is the distance threshold.

[0041] Step 3: Alignment of tooth root fillet features: Solve for the optimal rigid body transformation within the fillet feature area to achieve the best match between the initial profile and the measured profile; High curvature feature points are extracted in the tooth root fillet region. SVD decomposition is used to calculate the rotation and translation matrix, transforming the measured point cloud to the reference coordinate system. This avoids deviations caused by the participation of the wear region in the calculation, eliminates scaling and mirroring, ensures geometric authenticity, and achieves millimeter-level alignment accuracy, laying the foundation for micro-wear quantization. In step three, the fillet feature region is identified based on curvature radius constraints. ,in, This represents the local radius of curvature of a point in the point cloud, in mm. Equivalent curvature expression: Where k represents the local curvature, defined as the reciprocal of the radius of curvature. ; Matching the initial profile with the measured profile: , Where R: 3×3 rotation matrix, describing the rotational transformation of the coordinate system; T: 3×1 translation vector, describing the translation transformation of the coordinate system; Initial outline point cloud China belongs to The k-th point in the region; : Measured profile point cloud Zhongyu The corresponding nearest point; Euclidean norm, representing the Euclidean distance between two points; Constraints: Ensure the orthogonality of the rotation matrices. The constraints ensure that R is a valid rotation matrix, removing scaling or mirror transformations, where... Let R be the transpose of the rotation matrix. I is a 3×3 identity matrix. det(R): Determinant of the rotation matrix R; Solving for optimal rigid body transformation: Solving for the translation vector of the rotation matrix using SVD decomposition. Point cloud centralization: First, subtract the centroid of each point set in the R16 region from the point sets of the two point clouds. , ; Constructing the covariance matrix: Calculate the covariance matrix between two centered point clouds: H: 3×3 covariance matrix, which contains the spatial relationship between the two point clouds; SVD Decomposition (Singular Value Decomposition): Singular value decomposition of the bicovariance matrix H: ,in, U: A 3×3 left singular vector matrix, an orthogonal matrix. Σ: A 3×3 diagonal matrix containing singular values. V: A 3×3 right singular vector matrix, an orthogonal matrix. : Transpose of matrix V; Calculate the optimal rotation matrix: Calculate the optimal rotation matrix using U and V: ; Calculate the translation vector: After obtaining the rotation matrix R, the translation vector T is calculated using the centroid relation: Rotation matrix: Translation vector: ; Aligned point cloud: The aligned point cloud is obtained by performing a rigid body transformation on the measured point cloud. .

[0042] Step 4: Normal wear calculation: After alignment, calculate the normal wear depth at each point; After alignment, for each reference point, the nearest measured point is searched along the tooth profile normal direction, and the distance between the two is calculated as the normal wear depth, corresponding to the material volume loss, with clear physical meaning. A high-resolution wear distribution map is generated point by point, which can identify local abnormal wear, such as uneven wear and pitting. In step four, after alignment is completed, the normal wear depth of each point is calculated. The normal vector is determined by the pressure angle α, which is a constant value. ; For each initial profile point In the aligned measured point cloud Find its nearest point The normal wear depth is: .

[0043] Step 5: Wear distribution analysis: Perform parametric analysis on the wear data, parameterize the tooth surface, and convert the tooth surface coordinates into dimensionless parameters; In step five, the wear data is parametrically analyzed, and the tooth surface is parameterized by converting the tooth surface coordinates into dimensionless parameters. This method converts tooth surface coordinates into dimensionless tooth height and tooth width parameters, establishes a distribution function for wear depth, and calculates statistics such as mean, maximum, and standard deviation. It supports standardized analysis of batch data, facilitates the establishment of a wear database, identifies wear uniformity, provides a basis for lubrication condition assessment, concisely describes the severity of wear, and is suitable for generating inspection reports. Tooth height direction: Tooth width direction ,in, u: Dimensionless tooth height parameter, range is , v: Dimensionless tooth width parameter, range is , : The y-coordinate of the tooth root position (lowest point). : The y-coordinate of the tooth tip (highest point). : The z-coordinate of the starting position of the tooth width : The z-coordinate of the end position of the tooth width; Establish the relationship between wear depth and location parameters: Wear distribution function: ; Average wear: ; Maximum wear: ; Standard deviation: ; Distributed along the tooth height direction: ,in, , for The number of points in the set, Δy: the width of the analysis window in the tooth height direction; Tooth width distribution: ,in, , for The number of points in the set, Δz: the width of the analysis window in the tooth width direction.

[0044] Step Six: Results Output and Visualization: The final output consists of quantified parameters and visualized results.

[0045] Calculate the overall average wear, coordinates of the maximum wear location, and wear uniformity index to generate a 3D heat map and cross-sectional curve. Support threshold-based automatic alarms, such as early warnings when average wear exceeds limits, to help engineers quickly locate abnormal wear areas, guide maintenance decisions, and facilitate archiving and comparison with historical trends. In step six, the final output quantified parameters are: Overall average wear: ; Maximum wear location: ; Wear uniformity index: , range ; Visual output: 3D wear thermogram: Color mapping is a color gradient; Cross-sectional distribution curve: The changing trend.

[0046] Components not described in detail in this article are existing technologies.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic detection device for rack rail wear, comprising a rack rail structure (200), a steel rail (300) and a trolley framework (400), characterized in that, Also comprising: A walking part mounted on the trolley frame (400) for driving the detection equipment to walk along the railway rail (300) and to realize braking and static holding on the slope; A tooth profile scanning device (600) mounted on the trolley frame (400), the tooth profile scanning device (600) moves transversely and vertically along the trolley frame (400) and rotates, for scanning the rack structure (200) at multiple distances, multiple positions and multiple angles to obtain the three-dimensional tooth profile and the tooth surface wear distribution of the rack structure (200); A tackiness magnet device (800) symmetrically mounted on the trolley frame (400) for detecting the slope of the rail (300) and adjusting the normal pressure between the walking part and the rail (300) according to the slope.

2. A rack rail wear automated detection apparatus according to claim 1, wherein The tooth profile scanning device (600) comprises: A transverse moving linear magnetic rail (610) mounted in parallel on the trolley frame (400), a transverse moving linear mover (620) being slidably connected to the transverse moving linear magnetic rail (610); An up-down linear magnetic rail (650) mounted on the transverse moving linear mover (620), an up-down linear mover (630) being mounted on the up-down linear magnetic rail (650); A bracket (640) mounted on the up-down linear mover (630); A scanning angle control cylinder (660) comprising a cylinder main body (661) and a first telescopic rod (662), the cylinder main body (661) being mounted on the bracket (640), the first telescopic rod (662) being mounted on the cylinder main body (661); A laser scanner (670) comprising a scanner main body (671), a scanning window (672), a first rotary joint (673) and a second rotary joint (674), the scanner main body (671) being rotatably connected to the cylinder main body (661) and the first telescopic rod (662) through the first rotary joint (673) and the second rotary joint (674), the scanning window (672) being formed in the scanner main body (671).

3. A rack tooth wear automatic detection method, using the rack tooth wear automatic detection device of claim 1 or 2, characterized in that, Comprising the following steps: Step one: laser scanning data acquisition: obtaining the three-dimensional coordinate points and wear distribution of the structure through the detection equipment; Step two: data preprocessing and noise reduction: removing random noise of the scanned noise and outlier data through filtering algorithm, and then removing outliers; Step three: alignment of tooth root fillet features: solving the optimal rigid transformation in the fillet feature area to best match the initial profile and the measured profile; Step four: normal wear calculation: after alignment, calculating the normal wear depth of each point; Step five: wear distribution analysis: parameterizing the wear data, parameterizing the tooth surface, and converting the tooth surface coordinates into dimensionless parameters; Step six: result output and visualization: finally outputting the results in the form of quantitative parameters and visual output.

4. A method for automatic detection of rack rail wear according to claim 3, characterized in that In step one, the structural contour point cloud in the initial state is denoted as , and the structural contour point cloud in the actual measurement state is denoted as .

5. The rack tooth wear automated detection method of claim 3, wherein, In step two, noise reduction: , wherein , filter parameters for the spatial and intensity domains, respectively; Outlier points: , Wherein, k is the neighborhood size, and threshold is the distance threshold.

6. The rack tooth wear automated detection method of claim 3, wherein, In step three, the fillet feature area is identified based on the curvature radius constraint: wherein, represents the local curvature radius of a point in the point cloud, with the unit of mm; Equivalent curvature expression: where k denotes the local curvature, defined as the inverse of the radius of curvature, ; The initial profile is matched to the measured profile: , Wherein, R: 3×3 rotation matrix, describing the rotation change of the coordinate system; T: 3×1 translation vector, describing the translation transformation of the coordinate system; : initial profile point cloud belonging to the kth point of the region; : measured profile point cloud : corresponding nearest point : corresponding nearest point : Euclidean norm, denoting the Euclidean distance between two points; Constraints: Ensure the orthogonality of the rotation matrix: The constraints ensure that R is a valid rotation matrix, removing scaling or mirroring transformations, where, is the transpose of the rotation matrix R, I is a 3×3 unit matrix, det(R): determinant of the rotation matrix R.

7. A method for automatic detection of rack rail wear according to claim 6, characterized in that In step three, the optimal rigid body transformation is solved: the translation vector of the rotation matrix is ​​solved by SVD decomposition. Point cloud centralization: First, subtract the centroid of each point set in the R16 region from the point sets of the two point clouds. ， ； Constructing the covariance matrix: Calculate the covariance matrix between two centered point clouds: where H: 3x3 covariance matrix contains the spatial relationship between two point clouds; SVD decomposition (singular value decomposition): Singular value decomposition of the bi-covariance matrix H: wherein, U: A 3×3 left singular vector matrix, an orthogonal matrix. Σ: A 3×3 diagonal matrix containing singular values. V: A 3×3 right singular vector matrix, an orthogonal matrix. : transpose of matrix V; Compute optimal rotation matrix: Compute optimal rotation matrix from U and V: ; Calculate the translation vector: After obtaining the rotation matrix R, the translation vector T is calculated using the centroid relation: Rotation matrix: , translation vector: ; Aligned point cloud: the measured point cloud is rigidly transformed to obtain an aligned point cloud: .

8. A method for automatic detection of rack rail wear according to claim 7, characterized in that In step four, after the alignment is completed, the normal wear depth of each point is calculated, the normal vector is determined by the pressure angle a, which is a constant value: ; For each initial profile point its nearest point in the aligned measured point cloud is found The normal wear depth is then .

9. The rack tooth wear automated detection method of claim 3, wherein, In step five, the wear data is parametrically analyzed, and the tooth surface is parameterized by converting the tooth surface coordinates into dimensionless parameters. tooth height direction: tooth width direction , in, u: dimensionless tooth height parameter, ranging from , v: dimensionless tooth width parameter, ranging from , : y coordinate of the root position (lowest point) : y coordinate of the tip position (highest point) : z coordinate of the start of the tooth width, : z coordinate of end of tooth width position Establishing the relationship between the depth of wear and the position parameter: wear distribution function: ; Average wear: ; maximum abrasion: ; Standard deviation: ; Distribution in the tooth height direction: wherein, , is the number of points in the set, Ay: the width of the analysis window in the tooth height direction; Distribution in the tooth width direction: wherein, , is the number of points in the set, Δz: width of the analysis window in the tooth width direction.

10. A method for automatic detection of rack rail wear according to claim 9, characterized in that In step six, the final output quantization parameters are: Overall average wear: ; Maximum wear position: ; Wear evenness index: , ranging from ; Visual output: Three-dimensional wear thermograph: Color mapping to color gradient; Cross-sectional profile curve: of the change trend.

Images