Self-piercing rivet joint quality detection method and device, terminal equipment and medium

Abstract

The invention provides a self-piercing rivet joint quality detection method and device, terminal equipment and a medium. The self-piercing rivet joint quality detection method comprises the steps of obtaining three-dimensional point cloud data of an internal structure of a self-piercing rivet joint; performing local neighborhood statistical analysis on the three-dimensional point cloud data, calculating a normal vector of each point, and evaluating the reliability of the normal vector according to the reliability evaluation index; carrying out overall segmentation on the three-dimensional point cloud by utilizing an improved region growing algorithm; identifying the rotating surface from the plurality of coarse segmentation blocks, projecting a normal vector of the rotating surface to a unified plane based on the geometric features of the axis, and performing fine segmentation on the rotating surface in the unified plane by using a standard region growing algorithm; performing three-dimensional surface reconstruction on the segmentation blocks to obtain a solid three-dimensional model of each component of the self-piercing rivet joint workpiece; and extracting internal forming geometric parameters of the self-piercing rivet joint workpiece, and comparing the internal forming geometric parameters with a preset qualified range threshold value so as to detect the quality of the self-piercing rivet joint. According to the invention, the precision of nondestructive quality detection of the self-piercing rivet joint can be improved.

Description

Technical Field

[0001] This invention belongs to the field of industrial non-destructive testing technology, specifically relating to a method, device, terminal equipment, and medium for quality testing of self-piercing riveting joints. Background Technology

[0002] Self-Piercing Riveting (SPR), as an advanced cold joining technology, is widely used in the manufacture of lightweight structures in the automotive, aerospace, and other fields. Its connection quality directly affects the safety and reliability of the overall structure. Currently, the industry relies heavily on destructive testing when evaluating the quality of SPR joints. This involves taking samples from the production line, cutting, embedding, polishing, and microscopically observing them to measure key geometric parameters such as undercut, rivet leg opening, and undercut thickness.

[0003] The main pain points of existing technologies are: 1. Destructive: The testing process is irreversible, leading to sample scrapping, high costs, and the inability to achieve full inspection of every joint.

[0004] 2. Lag: The time from sampling to obtaining results is long, making it impossible to provide real-time feedback and guide production adjustments, which poses a quality risk.

[0005] 3. Subjectivity: Manual measurement is prone to error, resulting in poor consistency and repeatability of results.

[0006] In recent years, industrial CT technology has made non-destructive testing of SPR joints possible, enabling the acquisition of three-dimensional structural information within the joint. However, directly processing the massive voxel data (three-dimensional pixels) generated by CT scans is inefficient, and automatically and accurately extracting specific geometric parameters from complex three-dimensional structures remains a technical challenge that has not yet been effectively solved. Existing methods based on ultrasound and other techniques are difficult to effectively detect the complex internal structure of joints, resulting in the inability to generate cross-sectional images of the joint and difficulty in accurately reconstructing the three-dimensional morphology, leading to poor universality. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a method, device, terminal equipment and medium for quality inspection of self-piercing riveting joints, so as to improve the accuracy of non-destructive quality inspection of self-piercing riveting joints.

[0008] In a first aspect, the present invention provides a method for quality inspection of self-piercing riveting joints, the method comprising the following steps: Acquire three-dimensional point cloud data revealing the internal structure of the self-piercing riveting joint; Local neighborhood statistical analysis is performed on 3D point cloud data to calculate the normal vector of each point in the 3D point cloud, and the reliability of the normal vector is evaluated according to the reliability evaluation index. The reliability evaluation index is obtained by performing eigenvalue decomposition on the covariance matrix of each point. The covariance matrix represents the density of each point and its neighboring points in each directional dimension. An improved region growing algorithm is used to segment the 3D point cloud as a whole, resulting in multiple coarse segmentation blocks. The improved region growing algorithm uses the points corresponding to the normal vectors that meet the preset reliability conditions as seed points, and anchors the normal vectors of the seed points to the reference normal vectors of the regions where the seed points are located, so as to separate coarse segmentation blocks with regular boundaries. The rotating surface is identified from multiple coarse segments. Based on the axial geometric features, the normal vector of the rotating surface is projected onto a uniform plane. Then, the rotating surface is finely segmented in the uniform plane using a standard region growing algorithm to obtain multiple fine segments. The three-dimensional surface of the segmented block is reconstructed to obtain the solid three-dimensional model of each component of the self-piercing riveting joint workpiece; The internal forming geometry parameters of the self-piercing riveting joint workpiece are extracted based on the solid 3D model, and the internal forming geometry parameters are compared with the preset qualified range threshold to detect the quality of the self-piercing riveting joint.

[0009] Optionally, perform local neighborhood statistical analysis on the 3D point cloud data to calculate the normal vector of each point in the 3D point cloud, including: For each point in the 3D point cloud data, the neighboring points of the point are retrieved through the KD index tree to construct a local neighborhood point set for the point. Calculate the covariance matrix of the local neighborhood point set, and perform eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues; The direction of the eigenvector corresponding to the smallest eigenvalue is taken as the direction of the normal vector at that point.

[0010] Optionally, the expression for the reliability assessment index is:

[0011] in, Indicates the use of evaluating the first The evaluation index for the reliability of the normal vector at each point. , This represents the total number of points in the 3D point cloud data. , Indicates by the first The covariance matrix of the local neighborhood points of each point The three eigenvalues ​​obtained from the decomposition , , Indicates the number of adjacent points. Indicates from the first The point points to its first... A vector of adjacent points Indicates the first The average position of the neighboring points of each point , Represents the centered vector, covariance matrix The eigenvalue decomposition process is expressed as , express The corresponding feature vector.

[0012] Optionally, the process for determining whether reliability meets preset conditions is as follows: Set the first threshold ,when At that time, determine the first If the reliability of the normal vector of the nth point meets the preset condition, then the point lies on a smooth surface; otherwise, determine the nth point. The reliability of the normal vector at a point does not meet the preset condition, and the point is located on a rotating surface.

[0013] Optionally, the improved region growing algorithm, when performing the growing operation, only allows adjacent points whose normal vectors are within the same reference normal vector at an angle less than a preset angle threshold to be incorporated into the current region.

[0014] Optionally, projecting the normal vectors of the surface of revolution onto a uniform plane based on the axial geometry includes: By utilizing the geometric properties of the rotating surface, the axial parameters of the rotating surface are solved by fitting using the Levenberg-Marquardt optimization algorithm; the axial parameters include the coordinates of the axial points and the axial direction vector. Using the Rodriguez rotation formula, construct a transformation matrix that rotates the normal vector of any point on the surface of rotation to be coplanar with the normal vector of the seed point. By applying the transformation matrix, all normal vectors of the rotating surface are projected onto a uniform plane.

[0015] Optionally, the internal forming geometry parameters include at least one of undercut, rivet foot opening, and bottom remaining thickness.

[0016] Secondly, the present invention provides a self-piercing riveting joint quality inspection device, comprising: The data acquisition module is used to acquire three-dimensional point cloud data that reveals the internal structure of the self-piercing riveting joint; The normal vector calculation module is used to perform local neighborhood statistical analysis on 3D point cloud data, calculate the normal vector of each point in the 3D point cloud, and evaluate the reliability of the normal vector according to the reliability evaluation index. The reliability evaluation index is obtained by performing eigenvalue decomposition on the covariance matrix of each point. The covariance matrix represents the density of each point and its neighboring points in each directional dimension. The point cloud coarse segmentation module is used to perform overall segmentation of the 3D point cloud using an improved region growing algorithm to obtain multiple coarse segmentation blocks. The improved region growing algorithm uses the points corresponding to the normal vectors that meet the preset reliability conditions as seed points, and anchors the normal vectors of the seed points to the reference normal vectors of the regions where the seed points are located, so as to separate coarse segmentation blocks with regular boundaries. The point cloud fine segmentation module is used to identify the rotating surface from multiple coarse segmentation blocks. Based on the axial geometric features, the normal vector of the rotating surface is projected onto a unified plane, and the rotating surface is finely segmented in the unified plane using a standard region growing algorithm to obtain multiple fine segmentation blocks. The point cloud reconstruction module is used to reconstruct the three-dimensional surface of the segmented blocks to obtain the solid three-dimensional model of each component of the self-piercing riveting joint workpiece. The quality inspection module is used to extract the internal forming geometry parameters of the self-piercing riveting joint workpiece based on the solid 3D model, and compare the internal forming geometry parameters with the preset qualified range threshold to detect the quality of the self-piercing riveting joint.

[0017] Thirdly, the present invention provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method.

[0018] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0019] The present invention has at least the following beneficial effects: Quality inspection is performed using 3D point cloud data reflecting the internal structure of self-piercing riveting joints, completely avoiding physical cutting and achieving non-destructive testing and zero scrap. The reliability of normal vectors is evaluated through covariance matrix eigenvalue decomposition, effectively identifying and eliminating metal artifacts and scattering noise common in industrial CT scans, preventing noise from misleading subsequent calculations and ensuring the accuracy of measurement data from the source. An improved region growing algorithm, by anchoring seed point normal vectors, eliminates the cumulative error and drift caused by dynamic normal vector updates in traditional algorithms, enabling it to keenly distinguish the minute boundaries of close contact between the rivet and the sheet metal, thus improving quality inspection accuracy. For the rivet leg, a key rotating surface, innovative "axis projection unification" is used to eliminate radial geometric deviations, solving the problem of incomplete segmentation of cylindrical point clouds, ensuring the integrity of the measurement basis for core indicators such as undercut, and further improving quality inspection accuracy. Attached Figure Description

[0020] The accompanying drawings are provided to further understand the technical solutions of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the technical solutions of the present invention, and do not constitute a limitation on the technical solutions of the present invention.

[0021] Figure 1 This is a flowchart of a self-piercing riveting joint quality inspection method in one embodiment of this application; Figure 2 This is a schematic diagram of the overall segmentation result of a 3D point cloud in one embodiment of this application; wherein, Figure 2 (a) is the cross-sectional diagram of the joint. Figure 2 (b) is a schematic diagram of rivet image segmentation. Figure 2 (c) is a schematic diagram of image segmentation on the upper plate. Figure 2 (d) is a schematic diagram of the lower plate image segmentation; Figure 3 This is a schematic diagram of the structure of the self-piercing riveting joint quality inspection device in one embodiment of this application; Figure 4 This is a schematic diagram of the structure of a terminal device in one embodiment of this application. Detailed Implementation

[0022] The technical solution of the present invention will now be described in detail and completely with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0023] In the description of this invention, it should be noted that the terms "upper", "lower", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0024] Example 1 This embodiment is applied to an online quality inspection scenario on an automobile body production line. The object to be inspected is a self-piercing riveting joint connecting dissimilar materials such as aluminum and steel.

[0025] like Figure 1 As shown, the self-piercing riveting joint quality inspection method provided by the present invention specifically includes steps 11 to 16.

[0026] Step 11: Obtain three-dimensional point cloud data revealing the internal structure of the self-piercing riveting joint.

[0027] Specifically, the self-piercing riveting joint workpiece on the vehicle body is first scanned locally using an industrial CT scanner to obtain three-dimensional voxel data of the workpiece.

[0028] Subsequently, the 3D voxel data was preprocessed to convert it into data containing only 3D voxels. The data consists of 3D point cloud data with orientation coordinate information. In one feasible implementation, preprocessing includes noise reduction and downsampling. Noise reduction is based on statistical outlier removal algorithms (such as statistical filtering) and density-based noise filtering algorithms (such as radius filtering) to remove outliers and noise points generated during the scanning process. Downsampling uses methods such as voxel mesh downsampling to reduce point cloud density and improve subsequent processing efficiency while preserving key geometric features of the joint (such as rivet leg contours and plate interlocking boundaries) to the maximum extent.

[0029] This step replaces the traditional cutting and sampling method and forms the basis for all subsequent non-destructive quality testing.

[0030] Step 12: Perform local neighborhood statistical analysis on the 3D point cloud data, calculate the normal vector of each point in the 3D point cloud, and evaluate the reliability of the normal vector according to the reliability evaluation index.

[0031] In this embodiment of the invention, the reliability evaluation index is obtained by performing eigenvalue decomposition on the covariance matrix of each point. The covariance matrix characterizes the density of each point with its neighbors in each directional dimension.

[0032] Specifically, the process of performing local neighborhood statistical analysis on 3D point cloud data and calculating the normal vector of each point in the 3D point cloud includes steps 12.1.1 to 12.1.3.

[0033] Step 12.1.1: For each point in the 3D point cloud data, retrieve the neighboring points of the point through the KD index tree and construct the local neighborhood point set of the point.

[0034] For example, for any point KD-trees can be used to identify the set of neighboring points. , , This indicates the number of adjacent points. The value can be adjusted according to actual needs. Indicates from the first The point points to its first... A vector of neighboring points.

[0035] Step 12.1.2: Calculate the covariance matrix of the local neighborhood point set, and perform eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues.

[0036] Specifically, through the calculation formula

[0037] Get the first The covariance matrix of the local neighborhood points of each point , Indicates the first The average position of the neighboring points of each point , Represents the centered vector, covariance matrix The eigenvalue decomposition process is expressed as , express The corresponding feature vector, , Indicates by the first The covariance matrix of the local neighborhood points of each point The three eigenvalues ​​obtained from the decomposition .

[0038] Step 12.1.3: Take the direction of the eigenvector corresponding to the smallest eigenvalue as the direction of the normal vector of that point.

[0039] Specifically, the minimum eigenvalue corresponding feature vector The direction is used as a point The direction of the normal vector. The direction of the normal vector represents the orientation of the surface of an object (e.g., the orientation of the internal curved surface of a self-piercing riveting joint).

[0040] In another feasible implementation, for any point in the point cloud You can also select its neighboring points Construct a local neighborhood. Fit an optimal plane using the least squares method, minimizing the sum of squared distances from all points in the neighborhood to this plane, and thus obtain the normal vector. The objective function can be expressed as:

[0041] in, Let represent the normal vector to be found. It represents the neighborhood centroid.

[0042] Traditional methods simply use the normal vector at this point. However, CT point clouds contain noise, and in curved regions (especially at high curvature), the normal vector direction estimated based on small neighborhoods is inherently unstable and unreliable. Directly using these unreliable normal vectors leads to numerous errors in subsequent segmentation algorithms. To address this issue, the method provided in this invention evaluates the reliability of the normal vector based on a reliability assessment metric.

[0043] Specifically, the expression for the reliability assessment index is:

[0044] in, Indicates the use of evaluating the first The reliability evaluation index of the normal vector at each point. The physical meaning of this index is the proportion of the minimum eigenvalue to the total change. The smaller the value, the flatter the local surface, and the more reliable the normal vector estimation.

[0045] In this embodiment of the invention, the reliability judgment process of the normal vector is as follows: Set the first threshold (e.g., 0.08), when At that time, determine the first If the reliability of the normal vector of the nth point meets the preset condition, then the point is located on a smooth surface, and its normal vector has high reliability; otherwise, determine the nth point. The reliability of the normal vector at a point does not meet the preset conditions. The point is located on a rotating surface (such as the root of the cylindrical surface of a rivet leg), and the reliability of its normal vector is low.

[0046] Step 13: Use the improved region growing algorithm to perform overall segmentation of the 3D point cloud to obtain multiple coarse segmentation blocks.

[0047] In this embodiment of the invention, the improved region growing algorithm uses the points corresponding to the normal vectors that meet the preset reliability conditions as seed points, and anchors the normal vectors of the seed points to the reference normal vectors of the regions where the seed points are located, so as to separate coarsely segmented blocks with regular boundaries. Specifically, when performing the growing operation, the improved region growing algorithm only allows adjacent points whose normal vectors and reference normal vectors have an angle less than a preset angle threshold to be incorporated into the current region. In one feasible implementation, in addition to the restriction that the angle between the normal vector and the reference normal vector is less than the preset angle threshold, the Euclidean distance between the adjacent points that can be incorporated into the current region and the seed point must also be less than a preset distance threshold.

[0048] Points that meet the criteria are added to the current region, and growth continues with these new points as the frontier until no more new points meet the criteria. At this point, a coarsely segmented block with regular boundaries is generated. Then, the next seed point is selected, and the above process is repeated until all seed points are assigned to a coarsely segmented block. Points with "low reliability" (mainly distributed on the rotating surface) may not be absorbed by any region at this stage, forming some scattered fragments.

[0049] This invention, by "anchoring the reference normal vector," forces points within a region to have similar normal directions, which perfectly aligns with the physical characteristics of "components on the same plane." Therefore, the algorithm can accurately segment the planar portions of the upper and lower plates into independent, smoothly bounded blocks, effectively preventing adhesion (undersegmentation) between different plates. For surfaces of rotation, this step is intentionally "shelved," and further processing will be performed on them in subsequent fine segmentation.

[0050] It is worth mentioning that the "anchoring reference normal vector" in the growth process changes the dynamic standard in the traditional algorithm where each growth point is compared with the normal vector of its direct neighbor. It forces the normal vector in the region to be consistent with a fixed and reliable reference direction. This part works in conjunction with the evaluation and selection of "reliable seed points" mentioned earlier to effectively suppress the irregular diffusion of the region caused by point cloud noise and continuous changes in surface normal vectors, resulting in coarse segmentation blocks with unusually regular boundaries. This is an effect that conventional methods cannot achieve.

[0051] Step 14: Identify the rotating surface from multiple coarse segments, project the normal vector of the rotating surface onto a unified plane based on the axial geometric features, and use a standard region growing algorithm to finely segment the rotating surface in the unified plane to obtain multiple fine segments.

[0052] In this embodiment of the invention, the process of projecting the normal vector of the rotating surface onto a uniform plane based on the axial geometric features includes steps 14.1 to 14.3.

[0053] Step 14.1: Utilize the geometric properties of the rotating surface and solve for the axis parameters of the rotating surface by fitting the Levenberg-Marquardt optimization algorithm.

[0054] In this embodiment of the invention, the axis parameters include the coordinates of the axis point and the axis direction vector.

[0055] Specifically, the normal vector at any point on an ideal cylindrical surface should intersect the cylinder's axis. Using the Levenberg-Marquardt nonlinear optimization algorithm, a spatial line is found that minimizes the sum of the distances from the backward extensions of all normal vectors to this line. This line is the axis of the surface of revolution.

[0056] Step 14.2: Using the Rodriguez rotation formula, construct a transformation matrix that rotates the normal vector of any point on the rotation surface to be coplanar with the normal vector of the seed point.

[0057] It should be noted that the normal vectors at different positions on the cylindrical surface point in all directions in three-dimensional space, making direct clustering or growth extremely difficult. In this embodiment of the invention, a seed point is selected from the set of points on the rotating surface. Its normal vector is The goal is to construct a rotational transformation that eliminates the divergence of the normal vector direction caused by the curvature of the cylindrical surface. Specifically, it calculates the normal vector direction at any point on the cylindrical surface. Azimuth relative to the axis and seed point The difference in azimuth Using Rodriguez's rotation formula, the rotation angle about the axis is constructed. Transformation matrix The purpose of this matrix is ​​to group the points... normal vector element Rotate in reverse back to the orientation plane where the seed point is located.

[0058] Step 14.3: Apply the transformation matrix to project all normal vectors of the rotating surface onto a uniform plane.

[0059] Specifically, the normal vector for all points on the rotating surface. Apply rotation transformation: All after transformation All of them will lie in the same plane perpendicular to the axis. At this point, the normal vectors that originally diverged radially in three-dimensional space will converge in direction after the transformation. That is: all All are approximately parallel to the seed point normal vector .

[0060] At this point, using a standard and simple region growing algorithm, based on... By growing the angle between them (the threshold can be set to 20 degrees), it is easy to aggregate all points of the entire rotating surface into the same region.

[0061] By employing a series of operations—"axis solving → normal vector projection"—this invention transforms a complex, anisotropic segmentation problem in three-dimensional space into a simple, isotropic clustering problem in a two-dimensional plane. This fundamentally overcomes the failure of general algorithms on rotating surfaces, enabling complete and accurate segmentation of high-curvature rotating components such as rivet legs.

[0062] It should be noted that when general segmentation algorithms fail on rotating surfaces (resulting in oversegmentation), the conventional approach for those skilled in the art is to try more complex clustering algorithms (such as spectral clustering) or introduce additional geometric constraints, but these are often computationally complex and yield unstable results. This invention takes a completely different approach, deeply exploring the inherent geometric characteristics of rotating surfaces: their surface normal vectors diverge in three-dimensional space, but under specific transformations around an axis, they should exhibit consistency. Therefore, this invention does not "hard-segment" in the complex original three-dimensional space, but creatively transforms the problem onto a unified two-dimensional plane by solving for the axis and using the Rodriguez formula for projection. On this plane, the originally divergent normal vectors become vectors with similar directions, at which point segmentation can be easily achieved by applying a simple standard region growing algorithm. This solution ingeniously transforms a challenging three-dimensional segmentation problem into a simple two-dimensional segmentation problem, demonstrating unconventional thinking and outstanding results.

[0063] Step 15: Perform three-dimensional surface reconstruction on the segmented blocks to obtain solid three-dimensional models of each component of the self-piercing riveting joint workpiece.

[0064] Before performing 3D surface reconstruction, the present invention further includes: determining adjacent segmented blocks based on distance conditions and calculating the curvature of each segmented block. Over-segmented blocks with similar curvature and adjacent segments are merged to obtain the final overall segmentation result. This accurately separates different components (such as rivets, upper plates, and lower plates) and cross-sectional views and surfaces with different features within the joint, obtaining cleanly segmented point cloud sets belonging to the rivets, upper plates, and lower plates respectively. In one feasible implementation, the overall segmentation result is as follows: Figure 2 As shown.

[0065] For each set, common point cloud reconstruction methods can be used to obtain solid 3D models of each component of the self-piercing riveting joint workpiece.

[0066] Step 16: Extract the internal forming geometry parameters of the self-piercing riveting joint workpiece based on the solid 3D model, and compare the internal forming geometry parameters with the preset qualified range threshold to detect the quality of the self-piercing riveting joint.

[0067] In this embodiment of the invention, the internal forming geometry parameters include at least one of the following: undercut, rivet foot opening, and bottom remaining thickness.

[0068] Specifically, firstly, principal component analysis (PCA) or least squares cylindrical fitting algorithm is used to determine the central axis of the reconstructed rivet model, which is defined as the Z-axis. A plane perpendicular to the Z-axis is then established to construct a standardized local coordinate system.

[0069] Subsequently, around the central axis Z, a series of virtual cross-sectional planes passing through the central axis are generated at preset angular steps (e.g., every 10 or 15 degrees). The intersection lines of each plane with the 3D models of the rivets, upper plate, and lower plate are calculated, thereby generating multiple sets of 2D contour curves.

[0070] Finally, for each set of two-dimensional contour curves, the internal forming geometry parameters are extracted based on computational geometry algorithms: Undercut: In the cross-sectional profile, identify the furthest point of the outer contour of the rivet leg and the intersection point where the upper sheet metal is embedded in the lower sheet metal. Calculate the projected distance between the two points in the direction perpendicular to the Z-axis. Since there are multiple slices, the final undercut is the minimum (representing the weakest link) or average of all slice measurements.

[0071] Rivet leg opening: Calculate the Euclidean distance between the three-dimensional coordinate points at the ends of the two rivet legs.

[0072] Remaining bottom thickness: Identify the corresponding point on the same straight line between the lowest point of the bottom profile of the rivet leg and the outer surface profile of the lower sheet material. Calculate the Euclidean distance between the two points.

[0073] In one feasible implementation, the process of comparing the internal forming geometry parameters with a preset acceptable range threshold is as follows: If all extracted internal forming geometry parameters are within the preset qualified threshold range, then the current self-piercing riveting joint is determined to be of qualified quality. If any parameter is outside the range, the current self-piercing riveting joint is deemed substandard. For example, insufficient undercut may be caused by inadequate interlocking or a puncture in the base plate.

[0074] In another feasible implementation, the variance of the slice data at each angle can also be calculated. If the variance exceeds the preset value, it is determined to be "asymmetric deformation of the rivet (riveting misalignment)," and an early warning is given even if the internal forming geometric parameter values ​​are qualified.

[0075] Example 2 This invention provides a self-piercing riveting joint quality inspection device, such as... Figure 3 As shown, the self-piercing riveting joint quality inspection device 300 includes: Data acquisition module 301 is used to acquire three-dimensional point cloud data that reveals the internal structure of the self-piercing riveting joint; The normal vector calculation module 302 is used to perform local neighborhood statistical analysis on the 3D point cloud data, calculate the normal vector of each point in the 3D point cloud, and evaluate the reliability of the normal vector according to the reliability evaluation index. The reliability evaluation index is obtained by performing eigenvalue decomposition on the covariance matrix of each point. The covariance matrix represents the density of each point and its neighboring points in each directional dimension. The point cloud coarse segmentation module 303 is used to perform overall segmentation of the three-dimensional point cloud using an improved region growing algorithm to obtain multiple coarse segmentation blocks. The improved region growing algorithm takes the point corresponding to the normal vector that meets the preset reliability conditions as the seed point, and anchors the normal vector of the seed point to the reference normal vector of the region where the seed point is located, so as to separate out the coarse segmentation blocks with regular boundaries. The point cloud fine segmentation module 304 is used to identify the rotating surface from multiple coarse segmentation blocks, project the normal vector of the rotating surface onto a unified plane based on the axial geometric features, and use a standard region growing algorithm to finely segment the rotating surface in the unified plane to obtain multiple fine segmentation blocks; The point cloud reconstruction module 305 is used to reconstruct the three-dimensional surface of the segmented block to obtain the solid three-dimensional model of each component of the self-piercing riveting joint workpiece. The quality inspection module 306 is used to extract the internal forming geometric parameters of the self-piercing riveting joint workpiece based on the solid three-dimensional model, and compare the internal forming geometric parameters with the preset qualified range threshold to detect the quality of the self-piercing riveting joint.

[0076] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. Their specific functions and technical effects can be found in the method embodiments section, and will not be repeated here. Those skilled in the art will understand that, for the sake of convenience and brevity, the division of the above-mentioned functional units and modules is only used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0077] like Figure 4 As shown, embodiments of the present invention provide a terminal device, such as... Figure 4 As shown, the terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 4The diagram shows only one processor, a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, wherein the processor D100 executes the computer program D102 to implement the steps in any of the above method embodiments.

[0078] Specifically, when the processor D100 executes the computer program D102, it acquires three-dimensional point cloud data revealing the internal structure of the self-piercing riveting joint; performs local neighborhood statistical analysis on the three-dimensional point cloud data, calculates the normal vector of each point in the three-dimensional point cloud, and evaluates the reliability of the normal vector according to the reliability evaluation index; uses an improved region growing algorithm to perform overall segmentation of the three-dimensional point cloud to obtain multiple coarse segmentation blocks; identifies the rotating surface from the multiple coarse segmentation blocks, projects the normal vector of the rotating surface onto a unified plane based on the axial geometric features, and uses a standard region growing algorithm to perform fine segmentation of the rotating surface in the unified plane to obtain multiple fine segmentation blocks; performs three-dimensional surface reconstruction on the segmentation blocks to obtain solid three-dimensional models of each component of the self-piercing riveting joint workpiece; extracts the internal forming geometric parameters of the self-piercing riveting joint workpiece based on the solid three-dimensional models, and compares the internal forming geometric parameters with a preset qualified range threshold to detect the quality of the self-piercing riveting joint. Among these methods, quality inspection is performed based on 3D point cloud data reflecting the internal structure of the self-piercing riveting joint, completely avoiding physical cutting and achieving non-destructive testing and zero scrap of the self-piercing riveting joint. The reliability of the normal vector is evaluated by eigenvalue decomposition of the covariance matrix, effectively identifying and eliminating metal artifacts and scattering noise common in industrial CT, preventing noise from misleading subsequent calculations and ensuring the accuracy of measurement data from the source. The improved region growing algorithm eliminates the cumulative error and drift caused by the dynamic update of the normal vector in the traditional algorithm by anchoring the seed point normal vector, and can keenly distinguish the tiny boundaries of the tight fit between the rivet and the plate, which is conducive to improving the accuracy of quality inspection. For the rivet leg, a key rotating surface, the innovative use of "axis projection unification" eliminates radial geometric deviation, solving the problem of the difficulty in completely segmenting the cylindrical surface point cloud, ensuring the integrity of the measurement basis of core indicators such as undercut, which is conducive to improving the accuracy of quality inspection.

[0079] The processor D100 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0080] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may be an external storage device of the terminal device D10, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal device D10. Furthermore, the memory D101 may include both internal and external storage units of the terminal device D10. The memory D101 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory D101 can also be used to temporarily store data that has been output or will be output.

[0081] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0082] This application provides a computer program product that, when run on a terminal device, enables the terminal device to implement the steps described in the above-described method embodiments.

[0083] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0084] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.

Claims

1. A method of self-piercing rivet joint quality detection, characterized by, include: Acquire three-dimensional point cloud data revealing the internal structure of the self-piercing riveting joint; Local neighborhood statistical analysis is performed on the three-dimensional point cloud data to calculate the normal vector of each point in the three-dimensional point cloud, and the reliability of the normal vector is evaluated according to the reliability evaluation index. The reliability evaluation index is obtained by performing eigenvalue decomposition on the covariance matrix of each point. The covariance matrix represents the density of each point and its neighboring points in each directional dimension. An improved region growing algorithm is used to segment the 3D point cloud as a whole to obtain multiple coarse segmentation blocks. The improved region growing algorithm uses the points corresponding to the normal vectors that meet the preset reliability conditions as seed points, and anchors the normal vectors of the seed points as the reference normal vectors of the regions where the seed points are located, so as to separate coarse segmentation blocks with regular boundaries. The rotating surface is identified from the multiple coarse segments. The normal vector of the rotating surface is projected onto a uniform plane based on the axial geometric features. The rotating surface is then finely segmented in the uniform plane using a standard region growing algorithm to obtain multiple fine segments. The three-dimensional surface of the segmented block is reconstructed to obtain the solid three-dimensional model of each component of the self-piercing riveting joint workpiece; The internal forming geometry parameters of the self-piercing riveting joint workpiece are extracted based on the solid 3D model, and the internal forming geometry parameters are compared with a preset qualified range threshold to detect the quality of the self-piercing riveting joint.

2. The method for quality inspection of self-piercing riveting joints according to claim 1, characterized in that, The step of performing local neighborhood statistical analysis on the 3D point cloud data and calculating the normal vector of each point in the 3D point cloud includes: For each point in the 3D point cloud data, the neighboring points of the point are retrieved through the KD index tree to construct a local neighborhood point set for the point; Calculate the covariance matrix of the local neighborhood point set, and perform eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues; The direction of the eigenvector corresponding to the smallest eigenvalue is taken as the direction of the normal vector at that point.

3. The method for quality inspection of self-piercing riveting joints according to claim 2, characterized in that, The expression for the reliability assessment index is: in, Indicates the use of evaluating the first The evaluation index for the reliability of the normal vector at each point. , This represents the total number of points in the 3D point cloud data. , Indicates by the first The covariance matrix of the local neighborhood points of each point The three eigenvalues ​​obtained from the decomposition , , Indicates the number of adjacent points. Indicates from the first The point points to its first... A vector of adjacent points Indicates the first The average position of the neighboring points of each point , Represents the centered vector, covariance matrix The eigenvalue decomposition process is expressed as , express The corresponding feature vector.

4. The method for quality inspection of self-piercing riveting joints according to claim 3, characterized in that, The process for determining whether the reliability meets the preset conditions is as follows: Set the first threshold ,when At that time, determine the first If the reliability of the normal vector of the nth point meets the preset condition, then the point lies on a smooth surface; otherwise, determine the nth point. The reliability of the normal vector at a point does not meet the preset condition, and the point is located on a rotating surface.

5. The method for quality inspection of self-piercing riveting joints according to claim 4, characterized in that, The improved region growing algorithm, when performing the growing operation, only allows adjacent points whose normal vectors are less than a preset angle threshold to be incorporated into the current region.

6. The method for quality inspection of self-piercing riveting joints according to claim 5, characterized in that, The step of projecting the normal vector of the rotating surface onto a uniform plane based on the axial geometric features includes: The axial parameters of the rotating surface are solved by fitting the Levenberg-Marquardt optimization algorithm using the geometric properties of the rotating surface; the axial parameters include the coordinates of the axial point and the axial direction vector. Using the Rodriguez rotation formula, construct a transformation matrix that rotates the normal vector of any point on the rotation surface to be coplanar with the normal vector of the seed point. The transformation matrix is ​​applied to project all normal vectors of the rotating surface onto the uniform plane.

7. The method for quality inspection of self-piercing riveting joints according to claim 6, characterized in that, The internal forming geometry parameters include at least one of the following: undercut, rivet foot opening, and bottom remaining thickness.

8. A self-piercing riveting joint quality inspection device, characterized in that, include: The data acquisition module is used to acquire three-dimensional point cloud data that reveals the internal structure of the self-piercing riveting joint; The normal vector calculation module is used to perform local neighborhood statistical analysis on the 3D point cloud data, calculate the normal vector of each point in the 3D point cloud, and evaluate the reliability of the normal vector according to the reliability evaluation index; the reliability evaluation index is obtained by performing eigenvalue decomposition on the covariance matrix of each point; the covariance matrix represents the density of each point and its neighboring points in each directional dimension. The point cloud coarse segmentation module is used to perform overall segmentation of the three-dimensional point cloud using an improved region growing algorithm to obtain multiple coarse segmentation blocks. The improved region growing algorithm takes the point corresponding to the normal vector that meets the preset reliability condition as the seed point, and anchors the normal vector of the seed point to the reference normal vector of the region where the seed point is located, so as to separate out the coarse segmentation blocks with regular boundaries. The point cloud fine segmentation module is used to identify the rotating surface from the multiple coarse segmentation blocks, project the normal vector of the rotating surface onto a unified plane based on the axial geometric features, and use a standard region growing algorithm to finely segment the rotating surface in the unified plane to obtain multiple fine segmentation blocks; The point cloud reconstruction module is used to reconstruct the three-dimensional surface of the segmented block to obtain the solid three-dimensional model of each component of the self-piercing riveting joint workpiece. The quality inspection module is used to extract the internal forming geometric parameters of the self-piercing riveting joint workpiece based on the solid three-dimensional model, and compare the internal forming geometric parameters with a preset qualified range threshold to detect the quality of the self-piercing riveting joint.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images