Automatic detection method for wear and deformation of brake disc

Abstract

This invention relates to an automatic detection method for brake disc wear and deformation, belonging to the field of vehicle component inspection technology, and is particularly applicable to brake discs of large EMU trains such as high-speed trains. The method includes: acquiring point cloud data of the contours of the friction surfaces on both sides and synchronous rotation angle signals collected by a data acquisition module during the rotation of the brake disc; preprocessing the original dataset, including point cloud stitching and eccentricity correction, to obtain a preprocessed point cloud whose geometric center coincides with the rotation axis; registering the preprocessed point cloud with a standard brake disc point cloud model, calculating the measured thickness in a cylindrical coordinate system and comparing it with the standard thickness to generate a wear distribution; calculating deformation indicators such as end face circular runout, flatness, and thickness change based on the point cloud; outputting and displaying the detection results; achieving simultaneous detection of full-surface wear distribution and multi-dimensional deformation in a single scan, with high efficiency, and can be seamlessly integrated into the EMU maintenance production line for online full inspection.

Description

Technical Field

[0001] This invention relates to the field of vehicle component testing technology, and more specifically, to an automatic method for detecting brake disc wear and deformation. Background Technology

[0002] The brake disc is a core safety component of the basic braking system of high-speed trains. It is typically made of cast steel or forged steel, with diameters ranging from 500mm to over 750mm and thicknesses typically between 40mm and 80mm. During braking, the brake disc converts enormous kinetic energy into heat through intense friction with the brake pads. The instantaneous temperature of the disc surface can reach hundreds of degrees Celsius, accompanied by rapid heating and cooling cycles. Under these harsh conditions, the friction surface of the brake disc gradually wears down, resulting in thinning of the disc surface and material loss. Simultaneously, due to uneven heat distribution and residual stress release, the disc surface is highly susceptible to plastic deformation such as thermal warping and wavy deformation. This manifests as excessive deformation indicators, including end face runout, flatness deviations, and changes in brake disc thickness.

[0003] In the routine maintenance of high-speed trains, the wear and deformation of the brake discs are the core criteria for determining whether they can continue to be used. If the wear exceeds the safety limit, the brake disc will be weak and there is a risk of breakage. If the deformation exceeds the tolerance, it will cause brake shudder, abnormal vibration and noise, which will seriously affect the smoothness of driving and braking safety. Therefore, it is essential to conduct accurate and comprehensive geometric condition inspection of the brake discs.

[0004] However, existing technologies face limitations in inspecting large brake discs for high-speed trains. Currently, some maintenance sites still use traditional manual measurement methods, such as using long-range vernier calipers or micrometers to measure the thickness of the disc surface locally, and manually checking the end face runout with a dial indicator. High-speed train brake discs are large and heavy, making manual rotation and point-by-point measurement extremely laborious and inefficient. Furthermore, the thickness can only be obtained at a few discrete points on the disc surface, failing to capture the continuous wear distribution across the entire friction surface, easily overlooking potential safety hazards such as uneven wear, thermal cracks, or dents. More importantly, manual operation makes it difficult to guarantee the accuracy and repeatability of readings, failing to meet the precision and reliability requirements of high-speed railways.

[0005] Some large-scale testing equipment, such as coordinate measuring machines (CMMs), while possessing high measurement accuracy, come with enormous costs, low efficiency, and stringent requirements for operating environment and personnel skills, making them completely unsuitable for the high-efficiency pace of high-speed train maintenance lines. Furthermore, if the single-point laser scanning solution used in the field of small car brake disc inspection is transplanted to large brake discs, the problems become even more pronounced: due to the large friction area of ​​the disc surface, completing a full-surface point-by-point scan takes an extremely long time; the data matching accuracy problem caused by step-by-step acquisition of data from both sides is significantly amplified on the scale of larger disc diameters, leading to increased calculation errors in wear and brake disc thickness variations, making it difficult to meet the dual requirements of maintenance procedures for measurement accuracy and efficiency.

[0006] Therefore, there is a need to provide an automatic detection method specifically designed for the characteristics of brake discs in large trains such as high-speed trains, in order to solve a series of technical problems in the existing technology in terms of detection efficiency, acquisition of full surface wear distribution, and simultaneous and precise measurement of multiple deformation indicators. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the present invention aims to provide an automatic detection method for brake disc wear and deformation, solving the problems of low efficiency, inability to fully reflect wear distribution, and inability to simultaneously measure end face runout and thickness changes in existing technologies using manual or single-point scanning methods.

[0008] The above-mentioned technical objective of the present invention is achieved through the following technical solution: an automatic detection method for brake disc wear and deformation, comprising the following steps, specifically:

[0009] Step S1: Obtain the original dataset: The original dataset includes: the contour point cloud data of the friction surfaces on both sides of the brake disc collected by the data acquisition module during the rotation of the brake disc, and the rotation angle signal recorded synchronously.

[0010] Step S2, Dataset Preprocessing: The original dataset is preprocessed, including stitching the contour point cloud data into a three-dimensional point cloud according to the rotation angle signal, and performing eccentricity correction so that the geometric center of the three-dimensional point cloud coincides with the preset rotation axis, thereby obtaining the preprocessed point cloud.

[0011] Step S3: Wear amount estimation: Register the preprocessed point cloud with the pre-established standard brake disc point cloud model, calculate the measured thickness for each grid cell in the cylindrical coordinate system, and compare it with the standard thickness to generate a wear amount distribution containing angle and radius position information;

[0012] Step S4, Deformation estimation: Based on the preprocessed point cloud, calculate at least one of the following: end face circular runout value, flatness, and brake disc thickness change of the brake disc friction surface, and then use the calculation result as the deformation index.

[0013] Step S5, Prediction Result Output: Output the wear distribution in step S3 and the deformation index in step S4.

[0014] Optionally, in step S1, the original dataset is acquired by the data acquisition module synchronously triggering the line laser profile sensors located on both sides of the brake disc at a fixed frequency during the process of the brake disc rotating at a constant speed for at least one revolution, and the rotation angle signal is synchronously output by the angle encoder; wherein, the measurement width of a single laser line covers the entire radial width of the brake disc friction surface.

[0015] Optionally, in step S2, the contour point cloud data is stitched into a three-dimensional point cloud based on the rotation angle signal. The specific operation process is as follows:

[0016] Step S2.1: Perform statistical filtering on each frame of contour data to remove outlier noise points that deviate from the local mean by more than a preset multiple;

[0017] Step S2.2: Based on the rotation angle signal, convert and stitch the filtered contour lines of each frame to a unified rectangular coordinate system to form a complete original three-dimensional point cloud on both sides of the brake disc.

[0018] Optionally, the eccentricity correction in step S2 is specifically performed as follows:

[0019] Step S2.3: Extract points from the original 3D point cloud on one side of the brake disc that meet the preset radial range conditions. The radial range corresponds to the outer cylindrical surface or the central hole cylindrical surface of the brake disc.

[0020] Step S2.4: Use the least squares method to fit the extracted points to a cylindrical surface to obtain the coordinates of the geometric center;

[0021] Step S2.5: Calculate the offset vector of the geometric center coordinates relative to the preset rotation axis;

[0022] Step S2.6: Translate the original three-dimensional point clouds on both sides of the brake disc in the opposite direction of the offset vector so that the geometric center of the point cloud coincides with the preset rotation axis, thereby obtaining the preprocessed point cloud.

[0023] Optionally, the specific operation process in step S3 is as follows, including:

[0024] Step S3.1: Use the iterative nearest point algorithm to register the left and right point clouds in the preprocessed point cloud with the corresponding left and right models in the standard brake disc point cloud model, respectively, to obtain the registered left and right point clouds.

[0025] Step S3.2: Transform the registered left point cloud and the registered right point cloud from the Cartesian coordinate system to the cylindrical coordinate system respectively;

[0026] Step S3.3: In the cylindrical coordinate system, divide the grid into grid cells with a preset radial and circumferential interval. Take the median value of the Z coordinate of the point cloud in each grid cell as the left and right heights of the grid cell.

[0027] Step S3.4: Calculate the measured thickness of each grid cell, where the measured thickness is equal to the height on the right side minus the height on the left side;

[0028] Step S3.5: Obtain the standard thickness at the corresponding grid position from the standard brake disc point cloud model, and calculate the wear amount. Specifically, the wear amount is the standard thickness minus the measured thickness.

[0029] Step S3.6: Generate the wear distribution based on the wear amount of all grid points.

[0030] Optionally, the preset radial interval in step S3.3 is 1 mm, and the preset circumferential interval is 1°.

[0031] Optionally, in step S4, the calculation of the end face circular runout value includes:

[0032] Step S4.1: On the registered point cloud of the friction surface on one side of the brake disc, select multiple different radii along the radial direction and extract the circumferential contour corresponding to each radius;

[0033] Step S4.2: For each radius, calculate the difference between the maximum and minimum values ​​of the axial coordinates of all points on the circumference;

[0034] Step S4.3: Take the maximum value of the difference among all radii as the end face circular runout value of the friction surface on that side.

[0035] Optionally, in step S4, the specific operation process for calculating the flatness is as follows:

[0036] Step S4.4: The registered point cloud of the friction surface on one side of the brake disc is fitted to a reference plane using the least squares method;

[0037] Step S4.5: Calculate the distance from all points in the point cloud of the friction surface on this side to the reference plane, and take the sum of the absolute values ​​of the maximum positive distance and the maximum negative distance as the flatness value of the friction surface on this side.

[0038] Optionally, in step S4, the change in brake disc thickness is the difference between the maximum and minimum values ​​of the measured thickness obtained in step S3 within the entire friction surface area.

[0039] Optionally, step S5 specifically includes:

[0040] Step S5.1: Compare the calculated wear distribution, end face runout value, flatness value, and brake disc thickness change with the preset wear threshold, runout threshold, flatness threshold, and thickness change threshold, respectively.

[0041] Step S5.2: Highlight the grid areas where the wear exceeds the wear threshold, and issue warning signs for each deformation index that exceeds the corresponding threshold.

[0042] In summary, the present invention has the following beneficial effects:

[0043] 1. The data acquisition module continuously collects contours during rotation, generating a high-density three-dimensional point cloud covering the entire friction surface. Wear distribution maps are generated at extremely fine intervals in both the radial and circumferential directions, elevating wear detection from limited discrete sampling points to a complete two-dimensional continuous distribution. Any local abnormal wear area can be accurately captured and located, thus providing complete and reliable data support for the safety assessment of the brake disc. Simultaneously, based on the registered point cloud and measured thickness data obtained from a single scan, wear distribution, end face runout value, flatness value, and brake disc thickness change can be calculated and output synchronously. This achieves comprehensive evaluation of multiple key indicators using source data, providing maintenance personnel with a comprehensive and intuitive decision-making basis for determining whether the brake disc can continue to be used. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the implementation process of the method of the present invention. Detailed Implementation

[0045] To make the objectives, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Several embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein.

[0046] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances. The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature.

[0047] In this invention, unless otherwise expressly specified and limited, "above" or "below" a second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of a second feature includes the first feature being directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" of a second feature includes the first feature being directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature. The terms "vertical," "horizontal," "left," "right," "above," "below," and similar expressions are for illustrative purposes only and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed or operated in a specific orientation, and therefore should not be construed as limiting the invention.

[0048] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0049] This invention provides an automatic detection method for brake disc wear and deformation, such as... Figure 1 As shown, it includes the following steps, specifically:

[0050] Step S1: Obtain the original dataset: The original dataset includes: the contour point cloud data of the friction surfaces on both sides of the brake disc collected by the data acquisition module during the rotation of the brake disc, and the rotation angle signal recorded synchronously.

[0051] Step S2, Dataset Preprocessing: The original dataset is preprocessed, including stitching the contour point cloud data into a three-dimensional point cloud according to the rotation angle signal, and performing eccentricity correction so that the geometric center of the three-dimensional point cloud coincides with the preset rotation axis, thereby obtaining the preprocessed point cloud.

[0052] Step S3: Wear amount estimation: Register the preprocessed point cloud with the pre-established standard brake disc point cloud model, calculate the measured thickness for each grid cell in the cylindrical coordinate system, and compare it with the standard thickness to generate a wear amount distribution containing angle and radius position information;

[0053] Step S4, Deformation estimation: Based on the preprocessed point cloud, calculate at least one of the following: end face circular runout value, flatness, and brake disc thickness change of the brake disc friction surface, and then use the calculation result as the deformation index.

[0054] Step S5, Prediction Result Output: Output the wear distribution in step S3 and the deformation index in step S4.

[0055] In a specific embodiment, a cast steel brake disc for a certain type of high-speed train is used as the test object. The outer diameter of the brake disc is 640mm, the inner diameter of the friction surface is 280mm, the radial width of the friction surface is about 180mm, the standard thickness is 45mm, and the material is heat-resistant cast steel. It should be noted that the scope of protection of the present invention is not limited to this specific size and model.

[0056] The detection system for implementing the automatic detection method used in this embodiment, in a preferred embodiment, includes a data acquisition module, a calculation and processing unit, and a display and output module;

[0057] The data acquisition module is the physical sensing and data acquisition part of the entire system. Its components include: a base; a rotating spindle, vertically mounted on the base via precision angular contact ball bearings; a pneumatic three-jaw automatic centering chuck, mounted on the top of the rotating spindle, used to coaxially clamp and fix the brake disc under test; a servo motor, connected to the rotating spindle via a flexible coupling, working with a servo driver to achieve precise closed-loop speed control; an angle encoder, coaxially mounted with the rotating spindle; in this embodiment, an incremental photoelectric encoder with a resolution of 0.005° is selected, with a maximum output frequency of 2MHz, capable of outputting orthogonal pulse signals precisely corresponding to the rotation angle in real time; and left and right laser profile sensors, symmetrically arranged on both sides of the brake disc via sensor mounting brackets with precision adjustment functions. The sensor mounting brackets are constructed of high-strength aluminum alloy profiles and equipped with precision micro-adjustment slides, allowing for precise adjustment of the sensor's pitch angle, yaw angle, and height.

[0058] The installation positions and orientations of the left and right laser profile sensors must simultaneously meet the following conditions: First, the planes on which the laser lines emitted by each sensor are located are parallel to the rotation axis of the main rotating shaft, to ensure that the laser lines are always projected radially onto the disc surface during the rotation of the brake disc; Second, the projection direction of the laser lines coincides with the radial direction of the brake disc, so that the profile data acquired by a single laser line accurately reflects the radial height change of the disc surface; Third, the effective measurement width of a single laser line is greater than the radial width of the brake disc friction surface.

[0059] In this embodiment, a CMOS linear laser profile sensor with an effective measurement width of 240 mm is selected. Its Z-axis repeatability is 0.2 μm, X-axis resolution is about 50 μm, profile point count is 1280 points / frame, and sampling frequency can reach up to 64 kHz, which can completely cover the entire radial range of the friction surface of the aforementioned 640 mm brake disc.

[0060] In this embodiment, the computing and processing unit is implemented using an industrial control computer, equipped with a multi-core high-performance processor, large-capacity memory, and solid-state drive storage. The industrial control computer internally integrates software programs for a point cloud preprocessing module, an eccentricity correction module, a point cloud registration module, a wear calculation module, and a deformation calculation module. Data flow between modules utilizes shared memory to ensure efficient processing of large amounts of point cloud data. The display and output module includes an industrial-grade high-resolution display and a database server, used to present the detection results and store historical data for querying.

[0061] Furthermore, the system can also be equipped with an automatic loading and unloading robot. The robot is connected to the computing unit via an industrial Ethernet protocol. The computing unit controls the robot to automatically load the brake discs to be tested and automatically unload and sort the tested brake discs according to preset detection cycle instructions, realizing unattended online fully automatic detection.

[0062] Before formally commencing testing, two preparatory tasks need to be completed sequentially: sensor calibration and the establishment of a standard brake disc point cloud model. Specifically:

[0063] Joint calibration of sensors: After installation, the left and right line laser profile sensors each have their own independent sensor measurement coordinate system; due to the unavoidable slight deviations in mechanical installation, there is a zero-position deviation on the Z-axis and a corresponding deviation on the X-axis along the laser line direction between the left and right sensor coordinate systems.

[0064] Z-axis zero-point deviation refers to the systematic difference in readings of two sensors in the thickness direction, while X-axis corresponding deviation refers to the fact that the height values ​​read by two sensors at the same physical position in the radial direction of the disk do not come from the same radius line. These two deviations need to be accurately calibrated and compensated to avoid errors in matching corresponding points of the point clouds on both sides in subsequent steps, distortion of the actual thickness calculation results, and in severe cases, even misjudging normal features of the disk as wear.

[0065] In this embodiment, a precision calibration block with known thickness and multiple stepped surfaces is used for joint calibration of the sensor. The calibration block is made of steel similar to that of the brake disc, and the thickness values ​​of each stepped surface are calibrated by a coordinate measuring machine with an uncertainty better than 1 μm.

[0066] The calibration process is as follows: Place the calibration block vertically at the center of the measurement area between the left and right sensors, and adjust the orientation of the calibration block so that the normal of its stepped surface is parallel to the laser emission direction of the sensor. Drive the sensors on both sides to scan the corresponding stepped surfaces on the calibration block, and collect the contour data of each stepped surface.

[0067] By comparing the measured thickness differences of each stepped surface by the sensors on both sides with the true values ​​of the known thickness differences of the calibration block, an overdetermined set of equations is established. The least squares method is used to solve the mapping function between the Z-axis zero offset and the X-axis direction of the coordinate systems of the left and right sensors. The calibration parameters are stored in the calculation and processing unit for subsequent calculation of the measured thickness. This calibration process can be performed periodically to suppress and reduce parameter changes caused by factors such as temperature drift of the sensors.

[0068] Establish a standard brake disc point cloud model: The standard brake disc point cloud model is the benchmark reference for calculating wear. It needs to be established before inspection and stored in the calculation processing unit for a long time. The specific establishment process is as follows:

[0069] Take a standard brake disc of the same model as the brake disc to be tested, which has been precision machined and whose dimensions have been confirmed to be qualified by the metrology department; install the standard brake disc on the rotary table, start the pneumatic chuck to clamp it, and ensure that the clamping state is consistent with that during normal testing.

[0070] Subsequently, following the same acquisition process as the formal test, the rotating spindle was driven to rotate the standard brake disc at a constant speed of 60 rpm for 360°, and the left and right sensors were simultaneously triggered to acquire contour data, record the rotation angle signal corresponding to each frame, and obtain the standard raw dataset.

[0071] The standard original dataset is subjected to the same preprocessing procedure as in subsequent step S2, including: performing statistical filtering on each frame of contour data to remove outlier noise points; converting and stitching the contour lines of each frame to a unified rectangular coordinate system based on the rotation angle signal to form complete original 3D point clouds on the left and right sides of the standard brake disc; extracting the point cloud of the outer cylindrical surface and performing least squares cylindrical fitting and eccentricity correction to make the geometric center of the point cloud coincide precisely with the rotation axis, thereby obtaining the standard point cloud;

[0072] Combining the Z-axis zero-position offset and X-axis mapping relationship obtained from the aforementioned sensor calibration, and the known precise thickness value of the standard brake disc confirmed by precision measuring tools, coordinate calibration and standard thickness matrix calculation are performed on the standard point cloud. Specifically, in a cylindrical coordinate system, a grid is divided with radial intervals of 1 mm and circumferential intervals of 1°. For each grid cell, the Z-coordinates of the left and right side point clouds are compensated according to the calibration parameters, and then the standard thickness is calculated. This process generates a standard thickness matrix that corresponds one-to-one with the grid positions. Finally, the calibrated standard point cloud and its corresponding standard thickness matrix are stored as a standard brake disc point cloud model.

[0073] During the formal testing process, after completing the above preparations, proceed in the following steps in sequence;

[0074] Step S1, Obtain the raw dataset: First, the operator manually or by an automated loading / unloading robot picks up the brake disc to be inspected from the inspection rack and accurately places it on the pneumatic three-jaw chuck. After receiving the clamping command, the three jaws of the chuck move synchronously towards the center to clamp the brake disc coaxially. The pressure of the pneumatic chuck can be adjusted according to the specifications of the brake disc through a proportional valve to ensure that the clamping force is sufficient to overcome the centrifugal force during rotation, while not causing clamping deformation to the brake disc.

[0075] The servo motor receives motion control commands and drives the rotating spindle to smoothly accelerate the brake disc from a stationary state to the set speed. In this embodiment, the set speed is 60 rpm. The reason for choosing this speed is that, on the one hand, it is necessary to ensure sufficient linear speed so that the line laser sensor can acquire clear contour data within a single frame exposure time, and on the other hand, it is necessary to control the time of one rotation within a reasonable range. At this speed, it takes 1 second to rotate 360°, which meets the requirements of subsequent detection efficiency.

[0076] Once the spindle speed stabilizes, the data acquisition module starts synchronous acquisition; the angle encoder continuously outputs incremental pulse signals that are strictly synchronized with the spindle rotation angle. The synchronous trigger circuit in the data acquisition module uses the angle encoder signal as a time reference to generate trigger pulses at a fixed frequency of 1000Hz, which are simultaneously sent to the left laser profile sensor and the right laser profile sensor.

[0077] Furthermore, at each frame trigger, the left and right sensors each complete a contour scan, each outputting contour line data distributed radially along the brake disc. The contour line data specifically includes the Z-axis height information and corresponding X-axis position information of a series of measurement points arranged sequentially along the laser line direction. The data acquisition module simultaneously latches the angle encoder count value at the current frame trigger moment. This count value is converted to obtain the precise rotation angle corresponding to the contour of that frame; thus, each frame of contour data is assigned a unique corresponding angle label.

[0078] The spindle continues to rotate until the cumulative rotation angle reaches 360°, completing a full scan, at which point data acquisition stops synchronously. After the entire rotation process is completed, the raw dataset obtained by the computing unit includes: the complete contour point cloud sequence collected by the left sensor, the complete contour point cloud sequence collected by the right sensor, and the rotation angle signal sequence corresponding to each frame of contour. In this embodiment, with a trigger frequency of 1000Hz and a rotation cycle of 1 second, each sensor collects 1000 frames of contour data, each frame contains 1280 measurement points, and the point cloud data on one side is approximately 1.28 million three-dimensional points, which can fully guarantee the surface information density requirements of subsequent calculations.

[0079] Step S2, Dataset Preprocessing: This step processes the raw dataset obtained in Step S1 step by step, sequentially performing noise removal, point cloud stitching, and eccentricity correction to obtain a high-precision point cloud with accurate geometric center and rotation axis alignment, ensuring reliable data quality for subsequent wear and deformation calculations.

[0080] First, statistical filtering is performed to remove noise. During the actual data acquisition process by the sensor, each frame of contour data on both sides inevitably contains outlier noise points caused by local reflections from the measured surface, obstruction by small foreign objects, interference from stray light from the environment, or electronic noise from the sensor. These noise points appear spatially as isolated points or small clusters of points that deviate significantly from the surrounding normal point cloud. If they are not filtered out, they will interfere with subsequent steps such as cylinder fitting, point cloud registration, and meshing calculations, leading to a reduction in measurement accuracy.

[0081] This embodiment employs an outlier filtering algorithm based on local neighborhood distance statistics. For each measurement point on the contour line of each frame, K points are selected to the left and right along the contour line direction, with K being 10 in this embodiment, resulting in a total of 21 points forming its local neighborhood. The average distance μ and standard deviation σ from the center point of all points within this neighborhood are calculated. If the distance of a point from the neighborhood mean exceeds a preset multiple η times the standard deviation, η is 2 in this embodiment, i.e., |di-μ|>2σ, then the point is identified as an outlier noise point and removed from the contour data of that frame. The adaptability of this filtering method lies in the fact that the threshold is dynamically adjusted according to the local data characteristics. It is more sensitive to filtering in flat areas of the disk surface, while automatically widening in areas with geometric abrupt changes such as edge steps, effectively protecting real geometric edge features from being misjudged as noise.

[0082] Next, point cloud stitching is performed. After statistical filtering, the contour data of each frame are still in the local sensor coordinate system with their respective sensors as the origin. All contours of the left sensor describe the radial height distribution of the left surface of the brake disc at their respective rotation angles, but their reference coordinate system orientations are different. Similarly, the right sensor is the same. Therefore, it is necessary to transform all contours to the same global space rectangular coordinate system based on the rotation angle information marked by each frame contour.

[0083] This embodiment establishes a global coordinate system as follows: the rotation axis of the main spindle is the Z-axis, the brake disc mounting plane is the XOY plane, and the encoder's zero-position direction at the start of data acquisition is the positive X-axis direction. For a rotation angle of... A frame of outline that describes the angle The radial cross-sectional shape of the disk surface at the location;

[0084] In the sensor's local coordinate system, the coordinates of a point on this frame are: ,in Let z be the radial position along the laser line, and z be the height value measured by the sensor. The transformation formula for converting this point to the global coordinate system is: ;

[0085] Applying the above transformation to all points in the frame completes the coordinate transformation of the frame's outline. Merging and superimposing the transformed outline points from both the left and right sides at all angles forms the complete original 3D point cloud of the left and right sides of the brake disc.

[0086] Finally, eccentricity correction is performed. Under ideal clamping conditions, the geometric center of the brake disc should be perfectly aligned with the rotation axis of the rotary table. However, in actual clamping operations, due to the combined effects of brake disc manufacturing tolerances, chuck clamping accuracy, and small foreign objects on the workpiece positioning surface, a slight translational offset inevitably exists between the geometric center of the brake disc and the rotation axis. If not corrected, this eccentricity will cause serious systematic errors in subsequent calculations. Furthermore, when calculating end face runout, eccentricity will cause the originally flat disc surface to appear as a false runout signal with sinusoidal fluctuations with the rotation angle in the data, resulting in a significantly larger runout measurement value. When calculating wear, eccentricity will cause misalignment of the correspondence between the left and right side point clouds, leading to deviations in thickness calculations.

[0087] The eccentricity correction algorithm in this embodiment is based on the outer cylindrical surface feature of the brake disc, and the specific process is as follows;

[0088] The first step is feature point extraction. In the original 3D point cloud on the left, points whose radial coordinates meet the preset radial range conditions are extracted. Since the outer diameter of the brake disc to be tested is known to be 640mm and the outer radius of the friction surface is about 320mm, all points whose radial coordinates are located within the annular band between 315mm and 325mm are extracted. The annular band exactly covers the outer cylindrical surface area of ​​the brake disc. The reason for choosing the outer cylindrical surface instead of the central hole cylindrical surface is that the outer cylindrical surface has a larger radius, and more points participate in the fitting under the same point cloud density, resulting in a more robust and robust fitting result.

[0089] The second step is least-squares cylindrical surface fitting. Let the equation of the ideal cylindrical surface be... ,in Let R be the coordinates of the intersection point of the cylindrical surface axis and the XOY plane, and R be the radius of the cylindrical surface; the extracted outer cylindrical surface point set... Substitute the values, construct the error function, and iteratively solve it using a nonlinear least squares optimization algorithm. The algorithm yields the optimal estimate of R; it exhibits good convergence and robustness, and can obtain reliable fitting results even if there are a few residual outliers in the point set.

[0090] The third step is to calculate the offset vector. The fitted vector is obtained... This refers to the coordinates of the brake disc's geometric center on the XOY plane. Since the origin (0,0) of the coordinate system is the position of the rotation axis, the eccentricity vector of the geometric center relative to the rotation axis is V = ;

[0091] The fourth step is point cloud translation correction. The X and Y coordinates of all points in both the original 3D point clouds on the left and right sides are subtracted by Cx and Cy respectively, meaning the entire point cloud is translated in the opposite direction of the offset vector V. After translation, the geometric center of the brake disc in the point cloud coincides with the origin, i.e., it precisely coincides with the axis of rotation.

[0092] The point cloud after eccentricity correction is called the preprocessed point cloud, which includes the preprocessed left point cloud and the preprocessed right point cloud. At this point, the error introduced by the clamping eccentricity has been suppressed, and all subsequent calculations are based on the ideal condition that the geometric center and the axis of rotation are almost aligned, thus ensuring measurement accuracy.

[0093] Step S3, Wear Prediction: This step is based on the preprocessed point cloud obtained in step S2 and the pre-established and stored standard brake disc point cloud model. Through point cloud registration, coordinate transformation, mesh generation and thickness comparison, a high-resolution wear distribution reflecting the wear of the brake disc friction surface is finally calculated and generated.

[0094] First, fine registration of the point cloud is performed. Although step S2 has completed the eccentricity correction, making the geometric center of the brake disc coincide with the axis of rotation, there may still be slight tilt or rotational deviation of the brake disc in the spatial coordinate system. These residual deviations come from two aspects: one is the slight tilt caused by the disc surface not being strictly perpendicular to the axis of rotation when the brake disc is clamped; the other is the rotational offset caused by the circumferential starting angle of the brake disc around the axis of rotation not being completely consistent with the standard model. These deviations need to be corrected to avoid positional mismatch between the standard thickness and the actual thickness of each mesh unit in the subsequent wear calculation.

[0095] In this embodiment, the iterative nearest point algorithm is used to complete the fine registration; the preprocessed left point cloud and the standard left point cloud in the standard brake disc point cloud model are used as one pair of inputs, and the preprocessed right point cloud and the standard right point cloud are used as another pair of inputs, and the registration operation is performed independently.

[0096] The basic process of the ICP algorithm is as follows: In each iteration, for each point in the source point cloud, search for the closest Euclidean distance point in the target point cloud to form a set of corresponding point pairs; based on all corresponding point pairs, solve a rigid body transformation matrix using singular value decomposition or quaternion method to minimize the sum of squared total distances between corresponding point pairs; apply this transformation matrix to the source point cloud; repeat the above process until convergence, i.e., the change in distance residual between two adjacent iterations is less than a preset threshold or the maximum number of iterations is reached. In this embodiment, the convergence threshold is set to 1×10. -6 mm, the maximum number of iterations is set to 50; the two sets of point clouds obtained after ICP registration are called the registered left point cloud and the registered right point cloud, respectively. They are spatially aligned precisely with the corresponding side of the standard model, providing a unified high-precision benchmark for subsequent thickness and deformation calculations.

[0097] Then, coordinate transformation and mesh generation are performed. The registered left and right point clouds are converted to Cartesian coordinate systems. Transform to cylindrical coordinate system The specific transformation relationship is as follows: The cylindrical coordinate system fits the geometric characteristics of rotating parts such as brake discs. In this coordinate system, the radial and circumferential positions of the disc surface are explicitly expressed, which facilitates subsequent wear analysis based on position.

[0098] In a cylindrical coordinate system, the friction surface area is divided into grids. In this embodiment, the radial interval is set to 1 mm, and the circumferential interval is set to 1°. For a 640 mm brake disc, the radial width of its friction surface is 180 mm, that is, from the inner diameter of 280 mm to the outer diameter of 640 mm. The inner radius of the friction surface is approximately 140 mm, and the outer radius is approximately 320 mm. Therefore, the radial area is divided into 180 grid intervals, and the circumferential 360° area is divided into 360 grid intervals.

[0099] The entire friction surface is divided into 180 × 360 = 64,800 grid cells, each with an area of ​​approximately several square millimeters. Within each grid cell, since the grid size is extremely small relative to the overall size of the brake disc, it can be reasonably approximated that the height of the friction surface within that grid is uniform.

[0100] For each grid cell, extract the Z coordinate values ​​of all points in the registered point cloud that fall within it, and take the median as the representative height of the grid cell; the height of the left grid cell is denoted as zL, and the height of the right grid cell is denoted as zR; the median is used here instead of the arithmetic mean because the median is much less sensitive to residual noise points and small outliers than the mean. Even if some noise points that are not completely filtered out are mixed in the grid, the median can still give a robust estimate of the height of the grid.

[0101] Next, the measured thickness and wear amount are calculated. For each grid cell on the friction surface, the formula for calculating the measured thickness tm is: tm = zR – zL; This subtraction operation is not simply subtracting the readings of the two sensors, but is based on calling the zero-position deviation compensation amount of the left and right sensors Z-axis pre-calibrated and stored in step S2: In the actual calculation, zR and zL are first compensated and corrected by the calibration parameters before the subtraction is performed to ensure that the two are calculated under a unified measurement reference.

[0102] From the pre-stored standard brake disc point cloud model, the standard thickness tstd at the corresponding position is extracted according to the coordinate (r, θ) index of the grid cell. The formula for calculating the wear amount Δw of the grid cell is: Δw = tstd – tm; A positive value of Δw indicates that there is material wear at this position, and the larger the value, the more severe the wear; Δw close to zero indicates that the thickness at this position is basically consistent with the standard state; A negative value of Δw indicates that the measured thickness at this position is greater than the standard thickness, which may be due to material accumulation, oxide scale adhesion, or measurement abnormalities, and needs to be judged manually or automatically in combination with other information.

[0103] Finally, a wear distribution is generated. Based on the wear calculation results of all 64,800 grid points on the friction surface, a three-dimensional dataset containing angular coordinates (θ), radius coordinates (r), and wear values ​​(Δw) is generated, i.e., the wear distribution. This wear distribution completely describes the continuous spatial variation of wear on the entire friction surface of the brake disc.

[0104] Furthermore, the wear distribution can be plotted as a two-dimensional pseudo-color image. In this image, the horizontal axis represents the circumferential angle θ, and the vertical axis represents the radial coordinate r. The color of each pixel is mapped according to a preset color level based on the wear amount at that grid point. For example, blue represents areas of slight wear, green represents areas of moderate wear, and yellow and red represent areas of severe wear. Simultaneously, the image is labeled with the specific value of the maximum wear amount and its angular and radial coordinates. By observing this image, the operator can clearly grasp the overall wear condition of the entire friction surface and quickly locate abnormal wear areas such as uneven wear and localized pits.

[0105] In step S4, deformation is estimated. This step, based on the spatially registered point cloud and measured thickness data obtained in step S3, further calculates various deformation indicators of the brake disc. In this embodiment, the deformation indicators include three items: end face circular runout, flatness value, and brake disc thickness change. It should be noted that, to improve calculation accuracy and logical consistency, the calculation of all deformation indicators in this embodiment uses the point cloud that has completed ICP fine registration in step S3 as input. At this point, the point cloud is precisely aligned with the standard model in space, and the geometric datum is unified and reliable. Therefore, the preprocessed point cloud from step S2 is no longer used.

[0106] Deformation index 1: End face circular runout value; End face circular runout is an important indicator for measuring the axial swing amplitude of the brake disc surface when it rotates, and it is strictly limited in the EMU maintenance procedures; Excessive runout will directly lead to periodic braking torque fluctuations and bogie vibration during braking;

[0107] The specific method for calculating end face circular runout in this embodiment is as follows:

[0108] Step 1: Selection of radial sampling radii. On the left registered point cloud, multiple sampling radius values are uniformly selected along the radial direction at an interval of 10 mm; for the range from approximately 140 mm of the inner edge radius to 320 mm of the outer edge radius of the friction surface, a total of 19 sampling radii are selected (specifically 140 mm, 150 mm, 160 mm, ..., 320 mm respectively); multiple radii are selected instead of only measuring one radius at the outermost edge, because the brake disc may have local waviness deformation in the middle area of the friction surface, and measuring only a single radius may miss deformation features located near the inner diameter;

[0109] Step 2: Extraction of circular contour. For each selected sampling radius ri, a tiny radial tolerance band is set centered on this radius. In this embodiment, the tolerance band width is set to ±0.5 mm, that is, all points whose radial coordinates satisfy ri-0.5mm<r<ri+0.5mm are extracted; these points spatially form a discrete point set on an approximate circle around the rotation axis with a radius of approximately ri, which is the circular contour corresponding to this radius;

[0110] Step 3: Calculation of runout value for single radius. For the circular contour point set corresponding to each sampling radius ri, search for the maximum value of the Z-axis coordinates of all points in the set and the minimum value , then the runout value at this radius ; this value characterizes the maximum fluctuation range of the axial height of the disc surface at this radius;

[0111] Step 4: Determination of full disc surface runout. Traverse all 19 sampling radii, take the maximum runout value among them, that is , as the total circular runout of this side friction surface; this strategy of taking the maximum value ensures that local deformation existing at any radius will be captured and reflected in the final result, avoiding the risk of missed detection that may be caused by single-radius measurement;

[0112] The total circular runout of the right friction surface is calculated independently following exactly the same steps as above. Furthermore, the total circular runout of the left and right sides can be evaluated independently respectively, and the larger value of the two sides can also be taken as the overall total circular runout of the brake disc.

[0113] Deformation index 2: Flatness. Flatness is a key indicator for measuring the overall flatness of the friction surface of a brake disc, reflecting whether the disc surface has overall warpage or local concave-convex deformation. Different from total circular runout which focuses on axial runout in rotating state, flatness reflects the inherent geometric flatness of the disc surface;

[0114] The specific calculation process of flatness in this embodiment is as follows:

[0115] The first step is reference plane fitting. Extract all points belonging to the friction surface region from the point cloud after registration on the left. Use the least squares method to fit these points to a reference plane, minimizing the sum of squared distances from all points to this plane. Let the plane equation be Ax + By + Cz + D = 0. Solve for the optimal values ​​of coefficients A, B, C, and D using singular value decomposition.

[0116] The second step is to calculate the point-to-surface distance; for each point in the friction surface point cloud... Calculate its directed distance to the fitted reference plane. ; The plus or minus sign indicates which side of the reference plane the point is located on;

[0117] The third step is to evaluate the flatness value. Among all the point-to-plane distances, find the maximum positive distance dmax, which is the furthest point above the reference plane, and the maximum negative distance dmin, which is the furthest point below the reference plane.

[0118] According to the minimum area method evaluation principle, the flatness value of this side friction surface This value represents the minimum distance between two parallel planes when all points on the friction surface are contained within them, which conforms to the standard definition of flatness. Furthermore, the flatness value of the right friction surface is calculated independently using the above method, and the flatness values ​​of the left and right sides can also be evaluated independently.

[0119] Step S5, output the estimated results; This step integrates, visualizes, and automatically judges all the calculation results from steps S3 and S4 to generate a complete detection report;

[0120] The calculation and processing unit plots the wear distribution generated in step S3 into an intuitive two-dimensional pseudo-color map, which is then presented to the operator via an industrial display. On one side of the pseudo-color map, different colors are labeled with their corresponding wear value ranges, and the specific value of the maximum wear, along with its angle and radius coordinates, are indicated in text, allowing the operator to accurately identify the location of the most severe wear. Simultaneously, the various deformation indicators calculated in step S4, including the runout value of the left end face, the runout value of the right end face, the flatness value of the left side, the flatness value of the right side, and the change in brake disc thickness, are clearly displayed in a list format in the data area of ​​the inspection report interface.

[0121] Furthermore, the calculation and processing unit automatically compares the various test results with preset maintenance thresholds to achieve automatic judgment of pass / fail. In this embodiment, referring to the EMU maintenance procedures, the following judgment thresholds are set: the maximum wear amount does not exceed 2mm, exceeding this value indicates that the brake disc thickness is close to or below the safety lower limit, and there is a risk of insufficient strength; the end face runout does not exceed 0.05mm, exceeding this value will cause perceptible bogie vibration during braking; the flatness does not exceed 0.03mm, exceeding this value indicates that there is significant warping of the disc surface; the brake disc thickness change does not exceed 0.01mm, exceeding this value will cause periodic fluctuations in braking torque.

[0122] For the wear distribution map, grid areas exceeding the wear threshold are automatically highlighted with a striking warning color, such as red, which contrasts sharply with the color gradation of the normal areas, allowing operators to instantly identify the location and extent of severely worn areas.

[0123] For each deformation index, if an index exceeds the corresponding threshold, the index value will be highlighted in red with a "non-compliant" warning label; if all indexes are within the threshold range, a green "compliant" label will be displayed; all test data, judgment results, test time, brake disc number and other information are automatically stored in the database, supporting historical data traceability query and statistical analysis according to various conditions.

[0124] This invention discloses an automatic detection method for brake disc wear and deformation. A data acquisition module continuously collects contour data during rotation, generating a high-density three-dimensional point cloud covering the entire friction surface. Wear distribution maps are generated at extremely fine intervals in both the radial and circumferential directions, elevating wear detection from limited discrete sampling points to a complete two-dimensional continuous distribution. Any localized abnormal wear area can be accurately captured and located, thus providing complete and reliable data support for brake disc safety assessment. Simultaneously, based on the registered point cloud and measured thickness data obtained from a single scan, wear distribution, end face runout, flatness value, and brake disc thickness change can be calculated and output synchronously. This achieves comprehensive evaluation of multiple key indicators using source data, providing maintenance personnel with a comprehensive and intuitive decision-making basis for determining whether the brake disc can continue to be used.

[0125] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An automatic detection method for brake disc wear and deformation, characterized in that, The steps include the following: Step S1: Obtain the original dataset: The original dataset includes: the contour point cloud data of the friction surfaces on both sides of the brake disc collected by the data acquisition module during the rotation of the brake disc, and the rotation angle signal recorded synchronously. Step S2, Dataset Preprocessing: The original dataset is preprocessed, including stitching the contour point cloud data into a three-dimensional point cloud according to the rotation angle signal, and performing eccentricity correction so that the geometric center of the three-dimensional point cloud coincides with the preset rotation axis, thereby obtaining the preprocessed point cloud. Step S3: Wear amount estimation: Register the preprocessed point cloud with the pre-established standard brake disc point cloud model, calculate the measured thickness for each grid cell in the cylindrical coordinate system, and compare it with the standard thickness to generate a wear amount distribution containing angle and radius position information; Step S4, Deformation estimation: Based on the preprocessed point cloud, calculate at least one of the following: end face circular runout value, flatness, and brake disc thickness change of the brake disc friction surface, and then use the calculation result as the deformation index. Step S5, Prediction Result Output: Output the wear distribution in step S3 and the deformation index in step S4.

2. The automatic detection method for brake disc wear and deformation according to claim 1, characterized in that, In step S1, the original dataset is acquired by the data acquisition module synchronously triggering the line laser profile sensors located on both sides of the brake disc at a fixed frequency during the process of the brake disc rotating at a constant speed for at least one revolution. The rotation angle signal is synchronously output by the angle encoder. The measurement width of a single laser line covers the entire radial width of the brake disc friction surface.

3. The automatic detection method for brake disc wear and deformation according to claim 2, characterized in that, The specific operation process of stitching the contour point cloud data into a three-dimensional point cloud based on the rotation angle signal in step S2 is as follows: Step S2.1: Perform statistical filtering on each frame of contour data to remove outlier noise points that deviate from the local mean by more than a preset multiple; Step S2.2: Based on the rotation angle signal, convert and stitch the filtered contour lines of each frame to a unified rectangular coordinate system to form a complete original three-dimensional point cloud on both sides of the brake disc.

4. The automatic detection method for brake disc wear and deformation according to claim 3, characterized in that, The eccentricity correction in step S2 is specifically performed as follows: Step S2.3: Extract points from the original 3D point cloud on one side of the brake disc that meet the preset radial range conditions. The radial range corresponds to the outer cylindrical surface or the central hole cylindrical surface of the brake disc. Step S2.4: Use the least squares method to fit the extracted points to a cylindrical surface to obtain the coordinates of the geometric center; Step S2.5: Calculate the offset vector of the geometric center coordinates relative to the preset rotation axis; Step S2.6: Translate the original three-dimensional point clouds on both sides of the brake disc in the opposite direction of the offset vector so that the geometric center of the point cloud coincides with the preset rotation axis, thereby obtaining the preprocessed point cloud.

5. The automatic detection method for brake disc wear and deformation according to claim 1, characterized in that, The specific operation process in step S3 is as follows, including: Step S3.1: Use the iterative nearest point algorithm to register the left and right point clouds in the preprocessed point cloud with the corresponding left and right models in the standard brake disc point cloud model, respectively, to obtain the registered left and right point clouds. Step S3.2: Transform the registered left point cloud and the registered right point cloud from the Cartesian coordinate system to the cylindrical coordinate system respectively; Step S3.3: In the cylindrical coordinate system, divide the grid into grid cells with a preset radial and circumferential interval. Take the median value of the Z coordinate of the point cloud in each grid cell as the left and right heights of the grid cell. Step S3.4: Calculate the measured thickness of each grid cell, where the measured thickness is equal to the height on the right side minus the height on the left side; Step S3.5: Obtain the standard thickness at the corresponding grid position from the standard brake disc point cloud model, and calculate the wear amount. Specifically, the wear amount is the standard thickness minus the measured thickness. Step S3.6: Generate the wear distribution based on the wear amount of all grid points.

6. The automatic detection method for brake disc wear and deformation according to claim 5, characterized in that, The preset radial interval in step S3.3 is 1 mm, and the preset circumferential interval is 1°.

7. The automatic detection method for brake disc wear and deformation according to claim 1, characterized in that, In step S4, the calculation of the end face circular runout value includes: Step S4.1: On the registered point cloud of the friction surface on one side of the brake disc, select multiple different radii along the radial direction and extract the circumferential contour corresponding to each radius; Step S4.2: For each radius, calculate the difference between the maximum and minimum values ​​of the axial coordinates of all points on the circumference; Step S4.3: Take the maximum value of the difference among all radii as the end face circular runout value of the friction surface on that side.

8. The automatic detection method for brake disc wear and deformation according to claim 1 or 7, characterized in that, In step S4, the specific operation process for calculating the flatness is as follows: Step S4.4: The registered point cloud of the friction surface on one side of the brake disc is fitted to a reference plane using the least squares method; Step S4.5: Calculate the distance from all points in the point cloud of the friction surface on this side to the reference plane, and take the sum of the absolute values ​​of the maximum positive distance and the maximum negative distance as the flatness value of the friction surface on this side.

9. An automatic detection method for brake disc wear and deformation according to claim 1 or 7, characterized in that, In step S4, the change in brake disc thickness is the difference between the maximum and minimum values ​​of the measured thickness obtained in step S3 within the entire friction surface area.

10. The automatic detection method for brake disc wear and deformation according to claim 1, characterized in that, Step S5 specifically includes: Step S5.1: Compare the calculated wear distribution, end face runout value, flatness value, and brake disc thickness change with the preset wear threshold, runout threshold, flatness threshold, and thickness change threshold, respectively. Step S5.2: Highlight the grid areas where the wear exceeds the wear threshold, and issue warning signs for each deformation index that exceeds the corresponding threshold.

Images