A brake pad steel back piece defect detection method

Abstract

The application discloses a brake pad steel back piece defect detection method, and relates to the technical field of industrial nondestructive testing, comprising determining a surface residual stress risk area based on the geometric design parameters and stamping process parameters of the brake pad steel back piece; after pose correction of the brake pad steel back piece, laser thickness scanning, three-dimensional optical surface scanning and multi-physical field nondestructive testing are synchronously performed to obtain thickness data, three-dimensional point cloud data and ultrasonic, eddy current and thermal imaging response signals respectively; by fusing the geometric design parameters, stamping process parameters and multi-modal detection data, an adaptive defect detection mechanism with the residual stress risk area as the core is constructed, high sensitivity and high robustness identification of the surface and internal defects of the brake pad steel back piece are realized, and the method not only dynamically generates the risk area by using the local curvature, hole edge distance and process force ratio, but also tightens the surface geometric tolerance and enhances the internal defect signal weight according to the risk area.

Description

Technical Field

[0001] This invention relates to the field of industrial non-destructive testing technology, and in particular to a method for detecting defects in brake pad steel backing components. Background Technology

[0002] Industrial non-destructive testing (NDT) technology refers to a series of technical means that, without damaging or affecting the performance and structural integrity of the object being tested, utilize physical or chemical methods such as sound, light, electricity, magnetism, and heat to detect, locate, quantify, and evaluate internal and surface defects in materials, components, or engineering structures. Its core objective is to achieve early detection and condition determination of defects while ensuring product quality, equipment safety, and operational reliability. It is widely used in key industrial fields such as aerospace, rail transportation, energy and power, petrochemicals, and intelligent manufacturing.

[0003] Traditional nondestructive testing typically relies on a single physical field, making it difficult to simultaneously and effectively detect internal defects of different types, depths, and shapes. This carries the risk of missed detections or misjudgments. Furthermore, areas of residual stress concentration are often high-risk zones for defect initiation and propagation. However, traditional testing methods apply the same testing processes and parameters to all areas, lacking the means to optimize testing sensitivity or increase testing coverage based on stress field distribution. This makes it difficult to detect defects in high-risk areas early, reducing the early identification capability of defects in critical components. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for detecting defects in brake pad steel backing components. This solves the problem that traditional non-destructive testing usually relies on a single physical field, making it difficult to simultaneously and effectively detect internal defects of different types, depths, and shapes, which carries the risk of missed detections or misjudgments. In addition, areas of residual stress concentration are often high-incidence areas for defect initiation and propagation, but traditional testing methods use the same testing process and parameters for all areas, lacking means to optimize testing sensitivity or increase testing coverage based on stress field distribution. This makes it difficult to detect defects in high-risk areas early, reducing the early identification capability of defects in critical parts.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for detecting defects in the steel backing component of brake pads, comprising:

[0008] Based on the geometric design parameters and stamping process parameters of the brake pad steel backing component, the risk area of ​​surface residual stress is determined.

[0009] After the pose correction of the brake pad steel backing component, laser thickness scanning, three-dimensional optical surface scanning and multi-physics non-destructive testing are performed simultaneously to obtain thickness data, three-dimensional point cloud data and ultrasonic, eddy current and thermal imaging response signals, respectively.

[0010] After registering the 3D point cloud data with the corresponding CAD model, a tightened geometric tolerance is applied to the identification of surface defects in combination with the residual stress risk area, and the presence of geometric deformation defects is determined based on the thickness data.

[0011] The confidence score of internal defects is calculated based on the response signals of ultrasound, eddy current and thermal imaging, and the response signals falling into the residual stress risk area are given enhanced weights to determine internal defects.

[0012] Among them, the residual stress risk area is generated by mapping the local curvature change rate, hole edge distance and stamping process parameters, and is used to dynamically adjust the surface defect identification tolerance and internal defect scoring weight to achieve adaptive enhanced detection of high-risk areas.

[0013] As a preferred embodiment of the brake pad steel backing defect detection method of the present invention, the specific steps for determining the risk area of ​​surface residual stress based on geometric design parameters and stamping process parameters are as follows:

[0014] Extract surface triangular mesh data from the 3D CAD model of the brake pad steel backing component;

[0015] Perform differential geometric analysis on the triangular mesh and calculate the standard deviation of the angle between the normals of adjacent faces at each vertex, which is used as the local rate of curvature change.

[0016] Identify all through-hole structures in the 3D CAD model, and for each point on the edge of the hole, calculate its spatial straight-line distance to the nearest free boundary;

[0017] Read the stamping force and blank holder force values ​​recorded in the current stamping batch;

[0018] A residual stress sensitivity mapping function is constructed, and the local curvature change rate, the reciprocal of the hole edge distance, and the ratio of punching force to blank holder force are weighted and summed to obtain the residual stress risk index corresponding to each grid vertex.

[0019] Areas with a residual stress risk index greater than a preset threshold are marked as surface residual stress risk areas.

[0020] As a preferred embodiment of the brake pad steel backing defect detection method of the present invention, the step of calibrating the weighting coefficient using historical stamping test data specifically includes:

[0021] Select no fewer than thirty samples of brake pad steel backing parts of the same material grade and complete the stamping forming under different combinations of stamping parameters;

[0022] For each group of samples, the surface residual stress field was measured using an X-ray diffractometer to obtain the actual residual stress amplitude at each measurement point;

[0023] Simultaneously extract the local curvature change rate, hole edge distance, and punching force and blank holder force values ​​of the CAD model corresponding to each group of samples during the actual stamping process;

[0024] A multiple linear regression model was constructed using the actual residual stress amplitude as the target variable and the local curvature change rate, the reciprocal of the hole edge distance, and the ratio of punching force to blank holder force as input features.

[0025] The least squares method is used to find the weighting coefficient combination that minimizes the sum of squared errors between the predicted residual stress and the measured residual stress.

[0026] The obtained weighting coefficients are solidified as parameters of the residual stress sensitivity mapping function and used for risk area determination in subsequent online detection.

[0027] As a preferred embodiment of the brake pad steel backing component defect detection method of the present invention, the specific steps of performing position and orientation correction on the brake pad steel backing component are as follows:

[0028] The workpiece to be inspected is mounted on a special fixture at the end of a six-degree-of-freedom robotic arm. Three non-collinearly arranged high-resolution industrial cameras are activated to simultaneously acquire images of three preset reference holes on the workpiece.

[0029] A subpixel-level edge detection algorithm is performed on each image to accurately locate the pixel coordinates of the reference hole center in the image plane;

[0030] Based on the pre-calibrated camera intrinsic and extrinsic parameter matrices, the pixel coordinates are back-projected onto a unified world coordinate system to obtain the spatial measured coordinates of the three reference holes.

[0031] Read the ideal spatial coordinates of the three reference holes from the CAD model of the brake pad steel backing;

[0032] Solve for the optimal rigid body transformation that minimizes the sum of squared Euclidean distances between the ideal coordinates after rotation and translation transformations and the measured coordinates. Control the robotic arm to execute this rigid body transformation so that the workpiece is precisely aligned in the detection coordinate system.

[0033] As a preferred embodiment of the brake pad steel backing defect detection method of the present invention, the specific steps of determining whether there are geometric deformation defects based on thickness data are as follows:

[0034] The multi-point laser displacement sensor array is driven to move at a constant speed along a preset trajectory on the surface of the steel backing component, and the thickness measurement values ​​at discrete positions are collected synchronously. The collected thickness measurement values ​​are arranged into a thickness sequence according to the sampling order.

[0035] Calculate the arithmetic mean of the thickness sequence, and calculate the sum of squares of the deviations of each measurement in the thickness sequence from the arithmetic mean. Then divide by the sequence length minus one to obtain the thickness variance.

[0036] The standard deviation of thickness is obtained by taking the square root of the thickness variance, and the dynamic thickness variation coefficient is obtained by dividing the standard deviation of thickness by the arithmetic mean of the thickness series.

[0037] Based on the material grade used in the current steel backing component, the corresponding adaptive threshold is retrieved from the pre-stored database. If the dynamic thickness variation coefficient is greater than the adaptive threshold, it is determined that there is a geometric deformation defect.

[0038] In a preferred embodiment of the brake pad steel backing defect detection method of the present invention, the adaptive threshold is negatively correlated with the material yield strength, and is achieved through a piecewise linear interpolation table. The construction process of the piecewise linear interpolation table includes:

[0039] Establish a material-process benchmark library and collect no less than ten commonly used grades of brake pad steel backing raw materials, with each material grade corresponding to a unique yield strength value;

[0040] On a standardized stamping production line, using the same molds, stamping speeds, and lubrication conditions, produce no fewer than fifty qualified steel backing samples for each material grade.

[0041] For each qualified sample, perform a high-density thickness scan, with the scan trajectory covering the entire surface of the steel backing component and the number of sampling points not less than two thousand, to obtain a thickness dataset.

[0042] Based on the thickness dataset, the dynamic thickness variation coefficient is calculated independently for each sample;

[0043] For all qualified samples of the same material grade, the distribution of their dynamic thickness variation coefficient is statistically analyzed. After outliers are removed using statistical methods, the 95th percentile is taken as the benchmark threshold for the material grade.

[0044] The yield strength value corresponding to each material grade is used as the horizontal axis variable, and the corresponding benchmark threshold is used as the vertical axis variable to form a set of discrete data points in a two-dimensional coordinate system.

[0045] A continuous mapping relationship is constructed between two adjacent data points using a linear interpolation method. That is, when any value between the minimum and maximum yield strength is input, the corresponding adaptive threshold can be output by finding two adjacent points and calculating them proportionally.

[0046] During the online inspection phase, the material grade information of the workpiece to be inspected is read, and its corresponding yield strength value is retrieved from the material database.

[0047] Substitute the yield strength value into the piecewise linear interpolation table to calculate the adaptive threshold for the current workpiece.

[0048] As a preferred embodiment of the brake pad steel backing defect detection method of the present invention, the specific steps of registering the three-dimensional point cloud data with the corresponding CAD model are as follows:

[0049] Outlier removal and voxel mesh downsampling are performed on the raw point cloud data output by the 3D optical scanning device.

[0050] Extract planar features, cylindrical surface features, and circular hole features from the processed point cloud;

[0051] Extract geometric features of the same type from the CAD model and record their theoretical positions and orientations;

[0052] Using geometric feature matching as the initial constraint, initialize the rotation matrix and translation vector between the point cloud and the model;

[0053] The iterative nearest point algorithm is executed to find the nearest corresponding point on the model for each point in the point cloud in each iteration;

[0054] Calculate the root mean square error between the current corresponding point pairs. If the root mean square error is less than the convergence threshold, stop the iteration and output the final rigid body transformation parameters. Otherwise, update the transformation parameters and continue iterating until the convergence condition is met.

[0055] After registration, the point cloud data and the CAD model are aligned at the millimeter level in space.

[0056] As a preferred embodiment of the brake pad steel backing defect detection method of the present invention, the specific steps of assigning enhanced weights to the response signal falling into the residual stress risk region are as follows:

[0057] The spatial Boolean intersection operation is performed between the spatial sampling point coordinate sets of ultrasonic phased array detection, eddy current detection and active thermal imaging detection and the residual stress risk area mask, respectively.

[0058] For sampling points where the intersection result is true, multiply the original amplitude of the corresponding physical response signal by an enhancement factor greater than one.

[0059] The weighted signal is input into the internal defect confidence scoring model;

[0060] The internal defect confidence scoring model adopts a weighted fusion strategy, which integrates the normalized response values ​​of three types of nondestructive testing signals;

[0061] The normalized response value is obtained by dividing the original signal amplitude by the average response value of qualified parts made of the same material under the same testing conditions.

[0062] The final confidence score is the weighted sum of the three weighted normalized response values. When the confidence score exceeds the judgment threshold, the conclusion that there is an internal defect is output.

[0063] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the brake pad steel backing defect detection method as described in the first aspect of the present invention.

[0064] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the brake pad steel backing defect detection method as described in the first aspect of the present invention.

[0065] The beneficial effects of this invention are as follows: By integrating geometric design parameters, stamping process parameters, and multimodal detection data, an adaptive defect detection mechanism with residual stress risk areas as the core is constructed, which realizes high sensitivity and high robustness in identifying surface and internal defects of brake pad steel backing parts. The method not only dynamically generates risk areas by utilizing local curvature, hole edge distance, and process force ratio, but also tightens surface geometric tolerance and enhances the signal weight of internal defects accordingly. At the same time, an adaptive thickness variation threshold based on material yield strength is introduced to effectively distinguish between process abnormalities and normal fluctuations caused by material properties, thereby improving detection accuracy and reducing the false judgment rate. Attached Figure Description

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart of the method for detecting defects in the steel backing component of brake pads in Example 1. Detailed Implementation

[0068] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0069] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0070] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0071] Example 1, referring to Figure 1 As one embodiment of the present invention, this embodiment provides a method for detecting defects in brake pad steel backing components, including:

[0072] S1. Based on the geometric design parameters and stamping process parameters of the brake pad steel backing, determine the risk area of ​​surface residual stress.

[0073] Furthermore, surface triangular mesh data was extracted from the 3D CAD model of the brake pad steel backing.

[0074] The 3D CAD model of the brake pad steel backing component is constructed in 3D CAD software using parametric modeling based on the actual product engineering drawings or geometric parameters obtained through surveying. This includes creating a sketch based on the steel backing component's outline, mounting hole positions, riveting hole distribution, reinforcing rib structure, and stamping process characteristics, and performing extrusion, shelling, arraying, and chamfering operations to generate a solid model consistent with the mass-produced part. The model is then exported in a standard format to extract surface triangular mesh data.

[0075] Differential geometric analysis is performed on the triangular mesh to calculate the standard deviation of the angle between the normals of adjacent faces at each vertex, which is used as the local curvature change rate. All through-hole structures in the 3D CAD model are identified, and the spatial straight-line distance from each point on the edge of the hole to the nearest free boundary is calculated. The punching force and blank holder force values ​​recorded in the current stamping batch are read. A residual stress sensitivity mapping function is constructed, which weights and sums the local curvature change rate, the reciprocal of the hole edge distance, and the ratio of punching force to blank holder force to obtain the residual stress risk index corresponding to each mesh vertex. Regions with residual stress risk index greater than a preset threshold are marked as surface residual stress risk regions.

[0076] The residual stress sensitivity mapping function is expressed as:

[0077] ;

[0078] In the formula, This indicates the residual stress risk index; The local rate of curvature change is calculated from the standard deviation of the included angle between the normals of the neighboring surfaces of the triangular mesh vertex. This represents the distance from the edge of the hole to the nearest free edge; This indicates the value of the punching force applied by the main punch during the stamping process; This indicates the value of the blank holder force applied by the blank holder ring; , , The weighting coefficients, calibrated using historical stamping test data, reflect the relative contributions of geometric features and process parameters to the residual stress distribution.

[0079] The weighting coefficients were calibrated using historical stamping test data. This included selecting no fewer than thirty samples of brake pad steel backing parts of the same material grade and completing stamping under different combinations of stamping parameters. For each sample, the residual stress field on the surface was measured using an X-ray diffractometer to obtain the actual residual stress amplitude at each measurement point. Simultaneously, the local curvature change rate, hole edge distance, and stamping force and blank holder force values ​​of the CAD model corresponding to each sample were extracted.

[0080] Using the actual residual stress amplitude as the target variable and the local curvature change rate, the reciprocal of the hole edge distance, and the ratio of punching force to blank holder force as input features, a multiple linear regression model is constructed. The least squares method is used to solve for the combination of weighting coefficients that minimizes the sum of squared errors between the predicted residual stress and the measured residual stress. The obtained weighting coefficients are solidified as parameters of the residual stress sensitivity mapping function for use in subsequent online detection to determine the risk area.

[0081] Among them, the historical test samples cover the material yield strength range, stamping speed range and die wear state to ensure that the weighting coefficient has process robustness.

[0082] The objective function expression for weight coefficient calibration is:

[0083] ;

[0084] In the formula, The total number of historical test samples; For the first The residual stress risk index predicted by the mapping function at each sample grid point; The residual stress amplitude at the corresponding point was obtained by X-ray diffraction (after normalization and used as the target variable). , , , These represent the local curvature change rate, hole edge distance, punching force, and blank holder force values ​​corresponding to this sample.

[0085] It should be noted that by coupling geometric design features with stamping process parameters and calibrating weight coefficients based on measured residual stress data, the constructed residual stress risk area identification method can accurately reflect the distribution pattern of high stress concentration areas in the actual forming process. This effectively avoids misjudgments caused by traditional experience judgment or single parameter evaluation, provides a priori guidance driven by physical mechanisms for subsequent defect detection, and significantly improves the reliability and engineering applicability of high-risk area identification.

[0086] In particular, when identifying areas of residual stress risk, it is not simply a matter of applying geometric or process parameters. Instead, it combines abrupt changes in structural shape, hole edge constraint effects, and force balance relationships during the stamping process to form a comprehensive judgment criterion that reflects the stress concentration trend in actual forming. By reverse-calibrating the weight of each factor through a large amount of measured data, this judgment method can adapt to different materials, mold conditions, and process fluctuations, avoiding reliance on manual experience or fixed rules. This allows for the accurate identification of the areas most likely to cause problems before inspection, providing targeted guidance for subsequent inspections.

[0087] S2. After the positional correction of the brake pad steel backing, laser thickness scanning, three-dimensional optical surface scanning and multi-physics non-destructive testing are performed simultaneously to obtain thickness data, three-dimensional point cloud data and ultrasonic, eddy current and thermal imaging response signals, respectively.

[0088] Furthermore, the workpiece to be inspected is mounted on a dedicated fixture at the end of a six-degree-of-freedom robotic arm. Three non-collinearly arranged high-resolution industrial cameras are activated to simultaneously acquire images of three preset reference holes on the workpiece. A sub-pixel-level edge detection algorithm is executed on each image to accurately locate the pixel coordinates of the reference hole center in the image plane. Based on the pre-calibrated camera intrinsic and extrinsic parameter matrices, the pixel coordinates are back-projected to a unified world coordinate system to obtain the spatial measured coordinates of the three reference holes. The ideal spatial coordinates of the corresponding three reference holes are read from the CAD model of the brake pad steel backing component. The optimal rigid body transformation is solved to minimize the sum of the squares of the Euclidean distances between the ideal coordinates after rotation and translation transformations and the measured coordinates. The robotic arm is then controlled to execute this rigid body transformation, enabling the workpiece to achieve precise alignment in the inspection coordinate system.

[0089] The optimization objective expression for the optimal rigid body transformation is:

[0090] ;

[0091] In the formula, It is a three-dimensional rotation matrix. It is a translation vector; For the first in the CAD model Ideal spatial coordinates of each reference hole; The first result obtained by backprojection through a multi-camera vision system Measured spatial coordinates of one reference hole; This represents the Euclidean norm.

[0092] Among them, the pose correction accuracy must meet the requirement that the registration error of subsequent laser scanning and optical imaging is less than the preset tolerance.

[0093] It should be noted that by adopting a multi-camera vision system combined with sub-pixel-level edge positioning and rigid body transformation optimization strategies, high-precision pose correction of the workpiece in the inspection space was achieved, ensuring the spatial consistency of subsequent laser scanning, optical imaging and multi-physics field detection data in a unified coordinate system. This laid the geometric foundation for multi-source information fusion and precise defect positioning, and reduced false alarms and missed detections caused by clamping deviations or posture errors.

[0094] Specifically, the key to pose correction lies in ensuring that the workpiece's spatial position is highly consistent within the inspection system. Since subsequent inspection methods require extremely high positional accuracy, even minute posture deviations can lead to data misalignment and affect defect identification. By using three non-collinear reference holes in conjunction with high-precision visual positioning and employing mathematical optimization methods to calculate the optimal adjustment amount, the robotic arm actively executes the correction action, ensuring that each inspection is performed under a unified spatial reference. This reduces reliance on manual clamping accuracy and improves the stability and repeatability of the entire system in actual production line environments.

[0095] S3. After registering the 3D point cloud data with the corresponding CAD model, apply a tightened geometric tolerance to the surface defect identification based on the residual stress risk area, and determine whether there are geometric deformation defects based on the thickness data.

[0096] Furthermore, outlier removal and voxel mesh downsampling are performed on the raw point cloud data output by the 3D optical scanning device. Planar features, cylindrical features, and circular hole features are extracted from the processed point cloud. Geometric features of the same type are extracted from the CAD model, and their theoretical positions and orientations are recorded. Using geometric feature matching as the initial constraint, the rotation matrix and translation vector between the point cloud and the model are initialized. An iterative nearest-point algorithm is executed to find the nearest corresponding point on the model for each point in the point cloud in each iteration. The root mean square error between the current corresponding point pairs is calculated. If the root mean square error is less than the convergence threshold, the iteration is stopped and the final rigid body transformation parameters are output. Otherwise, the transformation parameters are updated and the iteration continues until the convergence condition is met. After registration, the point cloud data and the CAD model are aligned at the millimeter level in space.

[0097] A multi-point laser displacement sensor array is driven to move at a constant speed along a preset trajectory on the surface of the steel backing component, synchronously collecting thickness measurements at discrete positions. The collected thickness measurements are arranged into a thickness sequence according to the sampling order. The arithmetic mean of the thickness sequence is calculated, and the sum of squares of the deviations of each measurement in the thickness sequence from the arithmetic mean is calculated and then divided by the sequence length minus one to obtain the thickness variance. The square root of the thickness variance is taken to obtain the thickness standard deviation, and the thickness standard deviation is divided by the arithmetic mean of the thickness sequence to obtain the dynamic thickness variation coefficient. According to the material grade used in the current steel backing component, the corresponding adaptive threshold is retrieved from the pre-stored database. If the dynamic thickness variation coefficient is greater than the adaptive threshold, it is determined that there is a geometric deformation defect.

[0098] The expression for calculating the dynamic thickness variation coefficient is as follows:

[0099] ;

[0100] In the formula, Indicates the dynamic thickness variation coefficient; This indicates the total number of thickness measurements; Indicates the first Thickness measurements at each location; The arithmetic mean of the thickness sequence, i.e. .

[0101] The adaptive threshold is negatively correlated with the material's yield strength, and this is achieved through a piecewise linear interpolation table. The construction process of the piecewise linear interpolation table includes establishing a material-process benchmark library, collecting no fewer than ten commonly used grades of brake pad steel backing material, each material grade corresponding to a unique yield strength value; producing no fewer than fifty qualified steel backing sample pieces for each material grade on a standardized stamping production line using the same mold, stamping speed, and lubrication conditions; performing high-density thickness scanning on each qualified sample, with the scanning trajectory covering the entire surface of the steel backing, and sampling points of no fewer than two thousand, to obtain a thickness dataset; and independently calculating the dynamic thickness variation coefficient for each sample based on the thickness dataset.

[0102] For all qualified samples of the same material grade, the distribution of their dynamic thickness variation coefficient is statistically analyzed. After outliers are removed using statistical methods, the 95th percentile is taken as the baseline threshold for the material grade. The yield strength value corresponding to each material grade is used as the horizontal axis variable, and the corresponding baseline threshold is used as the vertical axis variable, forming a set of discrete data points in a two-dimensional coordinate system. A continuous mapping relationship is constructed between two adjacent data points using a linear interpolation method. That is, when any value between the minimum and maximum yield strength is input, the corresponding adaptive threshold can be output by finding two adjacent points and calculating proportionally. In the online inspection stage, the material grade information of the current workpiece to be inspected is read, and its corresponding yield strength value is retrieved from the material database. The yield strength value is substituted into the piecewise linear interpolation table to calculate the adaptive threshold of the current workpiece. The dynamic thickness variation coefficient is compared with the adaptive threshold. If the former is greater than the latter, a geometric deformation defect is determined to exist.

[0103] It should be noted that feature-guided iterative registration of 3D point clouds and CAD models, combined with dynamic adjustment of thickness variation criteria based on material properties, enables the determination of geometric deformation defects to have both high spatial alignment accuracy and adaptability to the process fluctuation characteristics of different materials. This effectively distinguishes between normal thickness changes caused by the intrinsic behavior of materials and real forming defects, improving the robustness and discrimination accuracy of the detection system, and is especially suitable for multi-material mixed-line production scenarios.

[0104] In particular, when judging geometric deformation defects, we not only focus on whether the thickness exceeds the tolerance, but also on the overall distribution characteristics of the thickness variation. By calculating the dispersion of the thickness data, we can identify potential problems where local fluctuations are within the limits but the overall distribution is uneven. At the same time, we set different judgment thresholds for different materials because high-strength materials naturally have smaller thickness fluctuations after forming, while ordinary materials allow for a certain range of fluctuations. The practice of dynamically adjusting the judgment criteria according to the material properties enables the system to capture real defects without misjudging normal process fluctuations as unqualified, thus improving the rationality and adaptability of the detection.

[0105] S4. Calculate the confidence score of internal defects based on the ultrasonic, eddy current and thermal imaging response signals, and assign enhanced weights to the response signals that fall into the residual stress risk area to determine internal defects.

[0106] Furthermore, the coordinate sets of spatial sampling points from ultrasonic phased array testing, eddy current testing, and active thermal imaging testing are respectively subjected to spatial Boolean intersection operations with the residual stress risk area mask. For sampling points with true intersection results, the original amplitude of their corresponding physical response signals is multiplied by an enhancement factor greater than one. The weighted signal is then input into the internal defect confidence scoring model. The internal defect confidence scoring model employs a weighted fusion strategy to synthesize the normalized response values ​​of the three types of nondestructive testing signals. The normalized response value is obtained by dividing the original signal amplitude by the average response value of qualified parts made of the same material under the same testing conditions.

[0107] The expression for calculating the normalized response value is:

[0108] ;

[0109] In the formula, , , These are ultrasound, eddy current, and thermal imaging at different locations. The original signal amplitude at that location; , , These are the average response values ​​of three types of signals measured on qualified workpieces of the same material grade under the same testing conditions.

[0110] The final confidence score is the weighted sum of the three weighted normalized response values. When the confidence score exceeds the judgment threshold, the conclusion that there is an internal defect is output.

[0111] The fusion expression for the internal defect confidence score is as follows:

[0112] ;

[0113] In the formula, Indicates spatial location Confidence score for internal defects at the location; , , These are ultrasound, eddy current, and thermal imaging at different locations. The normalized response value at the specified point is calculated as before; weights , , For position-adaptive weights, defined as if Located in the residual stress risk area Inside, then , , ;like If not within the residual stress risk area, then , , ;in, , , The basic weights for the three types of nondestructive testing methods satisfy the following conditions: ; This is a preset enhancement factor used to improve the sensitivity of defect signal discrimination in high-risk areas.

[0114] when Exceeding the preset judgment threshold At that time, it was determined that there was an internal defect at that location.

[0115] It should be noted that by spatially correlating multiphysics nondestructive testing signals with residual stress risk areas and implementing enhanced weighted fusion of responses in high-risk areas, the confidence score of internal defects is more focused on key parts that are prone to crack initiation or propagation. This not only improves the detection sensitivity of weak defect signals but also reduces the impact of noise interference in non-critical areas on the overall judgment, thereby achieving the forward-looking identification of potential structural failure hazards.

[0116] In particular, the determination of internal defects fully considers the importance of the defect's location. The same signal strength is more dangerous in areas of high residual stress than in ordinary areas; therefore, when fusing ultrasonic, eddy current, and thermal imaging signals, they are not treated equally, but signals located in risk areas are given higher priority. This approach simulates the judgment logic of engineers focusing on weak points, enabling the system not only to detect defects but also to assess their severity, thereby more effectively identifying internal problems that may evolve into potential failure hazards during use.

[0117] This embodiment also provides a computer device applicable to the brake pad steel backing component defect detection method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the brake pad steel backing component defect detection method proposed in the above embodiment.

[0118] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0119] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for detecting defects in the steel backing component of brake pads as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0120] In summary, this invention integrates geometric design parameters, stamping process parameters, and multimodal detection data to construct an adaptive defect detection mechanism centered on residual stress risk areas. This mechanism achieves highly sensitive and robust identification of surface and internal defects in brake pad steel backing components. The method not only dynamically generates risk areas using local curvature, hole edge distance, and process force ratio, but also tightens surface geometric tolerances and enhances the signal weight of internal defects accordingly. Furthermore, it introduces an adaptive thickness variation threshold based on material yield strength to effectively distinguish between process anomalies and normal fluctuations caused by material properties, thereby improving detection accuracy and reducing the false positive rate.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting defects in the steel backing component of brake pads, characterized in that: include: Based on the geometric design parameters and stamping process parameters of the brake pad steel backing, the risk area of ​​surface residual stress is determined. After the pose correction of the brake pad steel backing component, laser thickness scanning, three-dimensional optical surface scanning and multi-physics non-destructive testing are performed simultaneously to obtain thickness data, three-dimensional point cloud data and ultrasonic, eddy current and thermal imaging response signals, respectively. After registering the 3D point cloud data with the corresponding CAD model, a tightened geometric tolerance is applied to the identification of surface defects in combination with the residual stress risk area, and the presence of geometric deformation defects is determined based on the thickness data. The confidence score of internal defects is calculated based on the response signals of ultrasound, eddy current and thermal imaging, and the response signals falling into the residual stress risk area are given enhanced weights to determine internal defects. Among them, the residual stress risk area is generated by mapping the local curvature change rate, hole edge distance and stamping process parameters, and is used to dynamically adjust the surface defect identification tolerance and internal defect scoring weight to achieve adaptive enhanced detection of high-risk areas.

2. The method for detecting defects in the steel backing component of brake pads as described in claim 1, characterized in that: The specific steps for determining the risk area of ​​surface residual stress based on geometric design parameters and stamping process parameters are as follows: Extract surface triangular mesh data from the 3D CAD model of the brake pad steel backing component; Perform differential geometric analysis on the triangular mesh and calculate the standard deviation of the angle between the normals of adjacent faces at each vertex, which is used as the local rate of curvature change. Identify all through-hole structures in the 3D CAD model, and for each point on the edge of the hole, calculate its spatial straight-line distance to the nearest free boundary; Read the stamping force and blank holder force values ​​recorded in the current stamping batch; A residual stress sensitivity mapping function is constructed, and the local curvature change rate, the reciprocal of the hole edge distance, and the ratio of punching force to blank holder force are weighted and summed to obtain the residual stress risk index corresponding to each grid vertex. Areas with a residual stress risk index greater than a preset threshold are marked as surface residual stress risk areas.

3. The method for detecting defects in the steel backing component of brake pads as described in claim 2, characterized in that: The specific steps for calibrating the weighting coefficients using historical stamping test data are as follows: Select no fewer than thirty samples of brake pad steel backing parts of the same material grade and complete the stamping forming under different combinations of stamping parameters; For each group of samples, the surface residual stress field was measured using an X-ray diffractometer to obtain the actual residual stress amplitude at each measurement point; Simultaneously extract the local curvature change rate, hole edge distance, and punching force and blank holder force values ​​of the CAD model corresponding to each group of samples during the actual stamping process; A multiple linear regression model was constructed using the actual residual stress amplitude as the target variable and the local curvature change rate, the reciprocal of the hole edge distance, and the ratio of punching force to blank holder force as input features. The least squares method is used to find the weighting coefficient combination that minimizes the sum of squared errors between the predicted residual stress and the measured residual stress. The obtained weighting coefficients are solidified as parameters of the residual stress sensitivity mapping function and used for risk area determination in subsequent online detection. Among them, the historical test samples cover the material yield strength range, stamping speed range and die wear state to ensure that the weighting coefficient has process robustness.

4. The method for detecting defects in the steel backing component of brake pads as described in claim 3, characterized in that: The specific steps for correcting the position and orientation of the brake pad steel backing are as follows: The workpiece to be inspected is mounted on a special fixture at the end of a six-degree-of-freedom robotic arm. Three non-collinearly arranged high-resolution industrial cameras are activated to simultaneously acquire images of three preset reference holes on the workpiece. A subpixel-level edge detection algorithm is performed on each image to accurately locate the pixel coordinates of the reference hole center in the image plane; Based on the pre-calibrated camera intrinsic and extrinsic parameter matrices, the pixel coordinates are back-projected onto a unified world coordinate system to obtain the spatial measured coordinates of the three reference holes. Read the ideal spatial coordinates of the three reference holes from the CAD model of the brake pad steel backing; Solve for the optimal rigid body transformation that minimizes the sum of squared Euclidean distances between the ideal coordinates after rotation and translation transformations and the measured coordinates. Control the robotic arm to execute this rigid body transformation so that the workpiece is precisely aligned in the detection coordinate system. Among them, the pose correction accuracy must meet the requirement that the registration error of subsequent laser scanning and optical imaging is less than the preset tolerance.

5. The method for detecting defects in the steel backing component of brake pads as described in claim 4, characterized in that: The specific steps for determining whether geometric deformation defects exist based on thickness data are as follows: The multi-point laser displacement sensor array is driven to move at a constant speed along a preset trajectory on the surface of the steel backing component, and the thickness measurement values ​​at discrete positions are collected synchronously. The collected thickness measurement values ​​are arranged into a thickness sequence according to the sampling order. Calculate the arithmetic mean of the thickness sequence, and calculate the sum of squares of the deviations of each measurement in the thickness sequence from the arithmetic mean. Then divide by the sequence length minus one to obtain the thickness variance. The standard deviation of thickness is obtained by taking the square root of the thickness variance, and the dynamic thickness variation coefficient is obtained by dividing the standard deviation of thickness by the arithmetic mean of the thickness series. Based on the material grade used in the current steel backing component, the corresponding adaptive threshold is retrieved from the pre-stored database. If the dynamic thickness variation coefficient is greater than the adaptive threshold, it is determined that there is a geometric deformation defect.

6. The method for detecting defects in the steel backing component of brake pads as described in claim 5, characterized in that: The adaptive threshold is negatively correlated with the material's yield strength and is achieved through a piecewise linear interpolation table. The construction process of the piecewise linear interpolation table includes: Establish a material-process benchmark library and collect no less than ten commonly used grades of brake pad steel backing raw materials, with each material grade corresponding to a unique yield strength value; On a standardized stamping production line, using the same molds, stamping speeds, and lubrication conditions, produce no fewer than fifty qualified steel backing samples for each material grade. For each qualified sample, perform a high-density thickness scan, with the scan trajectory covering the entire surface of the steel backing component and the number of sampling points not less than two thousand, to obtain a thickness dataset. Based on the thickness dataset, the dynamic thickness variation coefficient is calculated independently for each sample; For all qualified samples of the same material grade, the distribution of their dynamic thickness variation coefficient is statistically analyzed. After outliers are removed using statistical methods, the 95th percentile is taken as the benchmark threshold for the material grade. The yield strength value corresponding to each material grade is used as the horizontal axis variable, and the corresponding benchmark threshold is used as the vertical axis variable to form a set of discrete data points in a two-dimensional coordinate system. A continuous mapping relationship is constructed between two adjacent data points using a linear interpolation method. That is, when any value between the minimum and maximum yield strength is input, the corresponding adaptive threshold can be output by finding two adjacent points and calculating them proportionally. During the online inspection phase, the material grade information of the workpiece to be inspected is read, and its corresponding yield strength value is retrieved from the material database. Substitute the yield strength value into the piecewise linear interpolation table to calculate the adaptive threshold for the current workpiece.

7. The method for detecting defects in the steel backing component of brake pads as described in claim 6, characterized in that: The specific steps for registering the 3D point cloud data with the corresponding CAD model are as follows: Outlier removal and voxel mesh downsampling are performed on the raw point cloud data output by the 3D optical scanning device. Extract planar features, cylindrical surface features, and circular hole features from the processed point cloud; Extract geometric features of the same type from the CAD model and record their theoretical positions and orientations; Using geometric feature matching as the initial constraint, initialize the rotation matrix and translation vector between the point cloud and the model; The iterative nearest point algorithm is executed to find the nearest corresponding point on the model for each point in the point cloud in each iteration; Calculate the root mean square error between the current corresponding point pairs. If the root mean square error is less than the convergence threshold, stop the iteration and output the final rigid body transformation parameters. Otherwise, update the transformation parameters and continue iterating until the convergence condition is met. After registration, the point cloud data and the CAD model are aligned at the millimeter level in space.

8. The method for detecting defects in the steel backing component of brake pads as described in claim 7, characterized in that: The specific steps for assigning enhanced weights to the response signals falling into the residual stress risk region are as follows: The spatial Boolean intersection operation is performed between the spatial sampling point coordinate sets of ultrasonic phased array detection, eddy current detection and active thermal imaging detection and the residual stress risk area mask, respectively. For sampling points where the intersection result is true, multiply the original amplitude of the corresponding physical response signal by an enhancement factor greater than one. The weighted signal is input into the internal defect confidence scoring model; The internal defect confidence scoring model adopts a weighted fusion strategy, which integrates the normalized response values ​​of three types of nondestructive testing signals; The normalized response value is obtained by dividing the original signal amplitude by the average response value of qualified parts made of the same material under the same testing conditions. The final confidence score is the weighted sum of the three weighted normalized response values. When the confidence score exceeds the judgment threshold, the conclusion that there is an internal defect is output.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the brake pad steel backing defect detection method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the brake pad steel backing defect detection method according to any one of claims 1 to 8.

Images