Anode guide rod center axis extraction method and device based on 3D point cloud and medium

Abstract

The application discloses an anode guide rod center axis extraction method and device based on a 3D point cloud, and a medium, and the method comprises the following steps: collecting point cloud data; preprocessing to obtain guide rod main body point cloud data; constructing a covariance matrix to solve eigenvalues and eigenvectors, and taking the eigenvector corresponding to the maximum eigenvalue as a rough extraction center axis direction vector; fitting an external cuboid through a minimum volume method to obtain a rough extraction center axis; dividing the guide rod main body point cloud data into several segments, and based on each segment of the guide rod main body point cloud data, obtaining each segment center axis through a segmented fitting method; and globally aligning to obtain an anode guide rod center axis. The application has the advantages that based on the point cloud data, noise background interference is effectively removed in the preprocessing stage, the data quality is improved, the global trend and local details are considered in the extraction process, the center axis extraction is realized through segmented fitting and global alignment, the geometric feature analysis and quantitative evaluation system are effectively integrated, and the application has the advantages of high precision, strong adaptability and high efficiency.

Description

Technical Field

[0001] This invention relates to the field of anode rod detection technology, specifically to a method, equipment, and medium for extracting the central axis of an anode rod based on 3D point clouds. Background Technology

[0002] As a key structural component in aluminum electrolysis production, the anode guide rod's geometry (especially its central axis, curvature, and eccentric displacement) directly affects the electrolytic cell's installation accuracy, current distribution, and operating energy consumption. Therefore, high-precision, multi-dimensional geometric inspection of the anode guide rod is crucial for ensuring stable operation of electrolysis production.

[0003] Currently, the main technical means in this field include mechanical contact measurement and surface analysis methods based on 3D scanning. Mechanical contact measurement methods, such as the one disclosed in Chinese patent CN222652056U, involve manually contacting the guide rod surface and reading the measurement value. This method relies on operator experience and can only measure the eccentric displacement of a local section of the guide rod, failing to obtain full-length bending information. Furthermore, manual operation is susceptible to interference from surface dirt and oxide layers, resulting in large repeatability errors (typically greater than ±0.5 mm), and measurement accuracy fluctuates with operator proficiency, making it difficult to meet the needs of industrial-scale batch testing.

[0004] Surface analysis methods based on 3D scanning, such as the method and related equipment for improving the performance of anode guide rods disclosed in Chinese patent CN120473000A, involve scanning the anode guide rod to construct a three-dimensional model, determining the proportion of protrusions on the surface of the anode guide rod based on the three-dimensional model, and then evaluating its conductivity. However, this method completely ignores the macroscopic geometric deformation of the guide rod (such as bending and axial offset), does not involve the extraction of the central axis and the calculation of curvature, and cannot effectively assess whether the overall structure of the guide rod meets the "straightness" requirements for electrolytic cell installation, thus creating a technological gap in structural integrity detection. Existing technologies for detecting anode guide rods have several technical shortcomings:

[0005] (1) The measurement dimension is singular, focusing only on the micro-features of local cross sections or surfaces, and lacks the ability to simultaneously acquire the entire axis of the guide rod, the overall curvature, and the local eccentric displacement.

[0006] (2) Low degree of automation: It relies on manual operation, which is inefficient, has poor repeatability, and is difficult to adapt to industrialized and large-scale testing scenarios.

[0007] (3) Weak anti-interference ability: It is easily affected by dirt, oxide layer, background noise and other interference on the guide rod surface, which affects the measurement accuracy.

[0008] (4) Lack of quantitative evaluation system: The lack of a scientific structural integrity evaluation standard based on geometric data has led to the easy flow of unqualified guide rods (such as excessive bending or excessive local eccentricity) into the production process, causing problems such as difficulty in electrolytic electrode hanging, current deviation, and increased energy consumption. Summary of the Invention

[0009] The technical problem to be solved by this invention is to provide an automatic extraction method, equipment and medium for the center axis of an anode guide rod based on 3D point cloud data. This method and medium can not only effectively remove noise and background interference and improve data quality in the preprocessing stage, but also take into account the global trend and local details in the axis extraction process. It can achieve high-precision center axis extraction through piecewise fitting and global iterative alignment, effectively integrate geometric feature analysis and quantitative evaluation system, and ensure the accuracy of anode guide rod detection, repeatability of results and standardization of evaluation. This method and medium has high precision, strong adaptability and high efficiency.

[0010] To address the aforementioned technical problems, the present invention provides a method for extracting the central axis of an anode guide rod based on 3D point clouds, comprising:

[0011] Collect 3D point cloud data of the anode guide rod;

[0012] The 3D point cloud data is preprocessed to remove noise points and background points, resulting in the main body point cloud data of the guide rod.

[0013] Based on the point cloud data of the guide rod body, a covariance matrix is ​​constructed and the eigenvalues ​​and corresponding eigenvectors are obtained. The eigenvector corresponding to the largest eigenvalue is used as the coarse extraction of the central axis direction vector.

[0014] Based on the coarse extraction of the central axis direction vector, the circumscribed cuboid is obtained by fitting using the minimum volume method, and the central axis of the circumscribed cuboid is extracted to obtain the coarse extraction of the central axis.

[0015] Based on the coarse extraction of the central axis direction, the point cloud data of the guide rod body is divided into several segments. Based on each segment of the guide rod body point cloud data, the central axis of each segment is obtained through a segmented fitting method.

[0016] Based on the coarse extraction of the central axis, each segment of the central axis is globally aligned to obtain the central axis of the anode guide rod.

[0017] In one optional implementation, it further includes:

[0018] Based on the central axis of the anode guide rod and the point cloud data of the guide rod body, the local eccentric displacement is calculated. Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis ;

[0019] Based on a preset threshold, through the local eccentric displacement Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis The calculation results are used to assess the structural integrity of the anode guide rod.

[0020] In one optional implementation, the preset threshold includes a first threshold. Second threshold and the third threshold ;

[0021] If the distance between the central axis of the anode guide rod and the theoretical axis Less than or equal to the first threshold The aforementioned local eccentric displacement Less than or equal to the second threshold and the overall curvature Less than or equal to the third threshold If the structure is intact, the anode guide rod is determined to be complete; otherwise, the anode guide rod is determined to be incomplete.

[0022] In one optional implementation, the preprocessing of the 3D point cloud data to remove noise and background points specifically includes the following steps:

[0023] Calculate the single-point deviation of the 3D point cloud data Based on a preset noise threshold For each data point in the 3D point cloud data, a determination is made; if the single-point deviation... Greater than the noise threshold If the data point is noisy, it will be deleted; otherwise, the data point will be retained.

[0024] Based on the geometric dimensions of the anode guide rod, the main body point cloud data of the guide rod is obtained by deleting background point data through an axisymmetric bounding box algorithm.

[0025] In one optional implementation, the preprocessing of the 3D point cloud data to remove noise and background points further includes the following steps:

[0026] The density of the 3D point cloud data is uniformized by using a voxel downsampling method.

[0027] Interpolation points are generated using a domain point cloud interpolation method to fill in the missing points in the 3D point cloud data;

[0028] If the cross-section of the anode guide rod end face is a standard rectangular cross-section, then based on the point cloud data of the rectangular cross-section, the coordinates of the four vertices of the rectangular cross-section (x1,y1,0), (x2,y1,0), (x2,y2,0) and (x1,y2,0) are determined by the least squares matrix fitting method. The center coordinates are determined as the origin of the coordinate system using the coordinates of the four vertices. A unified measurement coordinate system is established with the extension direction of the anode guide rod as the Z-axis, the long side direction of the rectangular cross-section as the X-axis, and the short side direction of the rectangular cross-section as the Z-axis.

[0029] If the cross-section of the anode guide rod end face is deformed, a segment of point cloud data is extracted along the extension direction of the anode guide rod. All point cloud data in this segment are projected and compressed onto the cross-section of the anode guide rod end face. The coordinates of the four vertices of the rectangular cross-section (x1,y1,0), (x2,y1,0), (x2,y2,0), and (x1,y2,0) are determined by fitting the data points after projection and compression. The center coordinates are determined as the origin of the coordinate system using the coordinates of the four vertices. A unified measurement coordinate system is established with the extension direction of the anode guide rod as the Z-axis, the long side direction of the rectangular cross-section as the X-axis, and the short side direction of the rectangular cross-section as the Y-axis.

[0030] In one optional implementation, based on the point cloud data of the guide rod body, a covariance matrix is ​​constructed and eigenvalues ​​and corresponding eigenvectors are obtained. The eigenvector corresponding to the largest eigenvalue is used as the coarse extraction center axis direction vector. Specifically, the following steps are included:

[0031] Based on the point cloud data of the guide rod body, the covariance matrix is ​​constructed, specifically using the following calculation method:

[0032] ,

[0033] The covariance matrix C satisfies symmetry. , , , This indicates the correlation between the X-coordinate set and the Y-coordinate set. This indicates the correlation between the X-coordinate set and the Z-coordinate set. This indicates the correlation between the Y-coordinate set and the Z-coordinate set. These represent the autocorrelation of the corresponding coordinate sets;

[0034] Based on the covariance matrix, the eigenvalues ​​and corresponding eigenvectors are solved using the following calculation method:

[0035] ,

[0036] in, It is a 3×3 identity matrix;

[0037] The solution obtained It includes three eigenvalues. and the The corresponding feature vector is the coarse extraction center axis direction vector. .

[0038] In one optional implementation, the step of dividing the guide rod body point cloud data into several segments based on the coarsely extracted central axis direction, and obtaining the central axis of each segment based on each segment of the guide rod body point cloud data through a segmented fitting method, specifically includes the following steps:

[0039] The anode guide rod is divided into N segments along the coarse extraction central axis, and the point cloud data of each segment is obtained. ,in Indicates the first The number of point cloud data for the main body of the guide rod;

[0040] Based on the point cloud data of each guide rod segment, a fitting error objective function is constructed. By minimizing the fitting error objective function, the local reference point and local direction vector of each segment are obtained. The specific calculation method is as follows:

[0041] ,

[0042] in, This is the local reference point for segment K, i.e. At the starting point of the central axis, This is the local direction vector of the Kth segment;

[0043] Based on each local reference point and local direction vector, the central axis of each segment is obtained, specifically using the following calculation method:

[0044] .

[0045] In one optional implementation, the step of globally aligning each segment of the center axis based on the coarsely extracted center axis to obtain the center axis of the anode guide rod specifically includes the following steps:

[0046] Data points on each segment of the central axis are sampled to obtain the source point set. ;

[0047] Data points on the coarse extraction center axis are sampled to obtain the target point set corresponding to the source point set. ;

[0048] Based on the source point set With the target point set The alignment error objective function is constructed and implemented using the following calculation method:

[0049] ,

[0050] in, For the source set Data points in the middle, For the target point set Zhongyu The corresponding data points For the source set The amount of data in the middle For rotation matrix, It is a translation matrix;

[0051] The alignment error objective function is solved iteratively until the iteration termination condition is met, and the rotation matrix is ​​obtained. Translation matrix ;

[0052] Based on the rotation matrix Translation matrix The center axis of each segment is corrected to obtain the center axis of the anode guide rod.

[0053] The present invention also provides a computer device, comprising: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, the steps of the method for extracting the central axis of an anode guide rod based on 3D point cloud as described in any of the preceding claims are performed.

[0054] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the method for extracting the central axis of an anode guide rod based on 3D point clouds as described in any of the preceding claims.

[0055] The method, equipment, and medium for extracting the central axis of the anode guide rod based on 3D point cloud of the present invention have the following advantages compared with the prior art:

[0056] (1) The method for extracting the central axis of the anode guide rod based on 3D point cloud of the present invention includes collecting 3D point cloud data of the anode guide rod. On the one hand, through non-contact 3D scanning, the geometric information of the entire length and all directions of the guide rod surface is obtained at one time, which lays the data foundation for the synchronous analysis of axis, curvature and eccentric displacement, and solves the problem of "single measurement dimension" in the existing technology. On the other hand, no manual contact and positioning are required, which significantly improves the detection efficiency and is suitable for industrial batch detection needs.

[0057] Preprocessing of 3D point cloud data removes noise and background points to obtain point cloud data of the guide rod body, ensuring a clean point cloud of the guide rod body and avoiding the introduction of noise and irrelevant points due to interference from surface dirt, oxide layer, and environmental background (such as tooling and ground). This provides high-quality data for subsequent analysis and ensures that the extracted center axis is not affected by the complex working conditions of the workshop.

[0058] Based on the point cloud data of the guide rod, a covariance matrix is ​​constructed and the eigenvalues ​​and corresponding eigenvectors are obtained. The eigenvector corresponding to the largest eigenvalue is used as the coarse extraction of the central axis direction vector, and the main direction of the overall extension of the guide rod is automatically calculated without the need for manual experience judgment. This ensures that the results are objective and repeatable, and provides a core and traceable directional benchmark for all subsequent quantitative calculations. It is a key link in realizing automation and quantitative evaluation.

[0059] Based on the coarse extraction of the central axis direction vector, a circumscribed cuboid is obtained by fitting using the minimum volume method. The central axis of the circumscribed cuboid is then extracted to obtain the coarse extraction of the central axis. A circumscribed cuboid that tightly encloses the point cloud is quickly fitted using the "minimum volume method." Its central axis can stably reflect the overall spatial position and orientation of the guide rod, is insensitive to local surface defects, and has strong robustness. It provides an excellent global initial value for subsequent refined piecewise fitting, avoiding the iterative algorithm from getting stuck in local optima or diverging due to poor initial values, thus ensuring the stability and efficiency of the entire process.

[0060] Based on the coarse extraction of the central axis direction, the point cloud data of the guide rod body is divided into several segments. Based on the point cloud data of each segment of the guide rod body, the central axis of each segment is obtained through a segmented fitting method. By using the segmented strategy, the global problem is decomposed into multiple local problems, and each segment of the point cloud is fitted independently. The resulting local axis can accurately reflect the subtle bending and offset of the guide rod segment, avoiding the limitation of existing technologies that can only measure individual sections. It realizes continuous and dense geometric feature sampling along the entire length of the guide rod, providing a direct and fine data source for calculating "overall bending degree" and "series of local eccentric displacements".

[0061] Based on the coarsely extracted center axis, each segment of the center axis is globally aligned to obtain the center axis of the anode guide rod. The fitted local axes of each segment, which may have slight misalignments, are then globally rotated and translated using the coarsely extracted axis as a reference to eliminate accumulated errors between segments. This ensures that the final spliced ​​center axis is a smooth, continuous, and highly accurate spatial curve. The finely extracted center axis is the gold standard for calculating the "overall average curvature" and any "local eccentric displacement of a cross-section," providing a direct, reliable, and authoritative final basis for subsequent threshold-based quantitative assessment of structural integrity (qualified / unqualified).

[0062] (2) The method for extracting the central axis of the anode guide rod based on 3D point cloud of the present invention further includes calculating the local eccentric displacement based on the central axis of the anode guide rod and the point cloud data of the guide rod body. Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis Based on a preset threshold, through local eccentric displacement Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis The calculation results are used to assess the structural integrity of the anode guide rod, including local eccentric displacement. It provides a comprehensive and objective data basis for assessing local deformation, including full-length curvature. This directly reflects the degree of "non-straightness" of the guide rod as a whole, and the distance between the central axis of the anode guide rod and the theoretical axis. This ensures that the baseline error of the entire process is controllable, thereby making the eccentric displacement calculated based on this axis... and overall curvature The results are reliable, providing a solid foundation for the final evaluation conclusion. On the other hand, by setting industrial thresholds, the above calculation results are automatically compared with the standards to establish an objective, unified, and executable quantitative evaluation standard, thereby achieving automatic decision-making. Attached Figure Description

[0063] Figure 1 This is a flowchart illustrating the extraction method of the anode guide rod center axis extraction method, equipment, and medium according to Embodiment 1 of the present invention.

[0064] Figure 2 This invention relates to a method, device, and medium for extracting the central axis of an anode guide rod based on 3D point cloud, specifically the acquisition of 3D point cloud data of the anode guide rod in Embodiment 1.

[0065] Figure 3 This is the point cloud data of the main body of the anode guide rod in Embodiment 1 of the method, equipment and medium for extracting the central axis of the anode guide rod based on 3D point cloud of the present invention;

[0066] Figure 4 This is a schematic diagram of the coarse extraction of the center axis of the anode guide rod based on 3D point cloud in Embodiment 1 of the present invention;

[0067] Figure 5 This is a schematic diagram of the anode guide rod center axis in Embodiment 1 of the method, equipment and medium for extracting the anode guide rod center axis based on 3D point cloud of the present invention. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0069] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "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.

[0070] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, an integral connection, or a detachable connection; they can refer to the internal connection of two components; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0071] Example 1

[0072] The method for extracting the central axis of the anode guide rod based on 3D point cloud in this embodiment is as follows: Figure 1 As shown, it includes:

[0073] Step S1. Acquire 3D point cloud data of the anode guide rod. For example... Figure 2 As shown, this embodiment uses a Faro Focus S70 3D laser scanner to acquire 3D point cloud data of the anode guide rod. The point cloud density is 100 points / mm², and the scanning range is 0.6-70m. On the one hand, through non-contact 3D scanning, the geometric information of the entire length and all directions of the guide rod surface can be acquired in one go, laying the data foundation for the simultaneous analysis of axis, curvature, and eccentric displacement, and solving the problem of "single measurement dimension" in existing technologies. On the other hand, no manual contact or positioning is required, which significantly improves the detection efficiency and is suitable for industrial batch detection needs.

[0074] Step S2. Preprocess the 3D point cloud data to remove noise and background points, obtaining the point cloud data of the guide rod body. For example... Figure 3 As shown, this ensures a clean point cloud of the guide rod body, avoiding the introduction of noise and irrelevant points due to surface dirt, oxide layers, and environmental background (such as tooling and ground). This provides high-quality data for subsequent analysis and ensures that the extracted central axis is not affected by the complex working conditions of the workshop. Step S2 specifically includes the following steps:

[0075] Step S21. Density homogenization of 3D point cloud data is performed using voxel downsampling. Point cloud density varies at different distances and angles. Voxel downsampling divides the 3D space into tiny cubes (voxels), retaining only one representative point (e.g., centroid) within each voxel. This efficiently homogenizes the spatial density of the entire point cloud. It avoids the dramatic increase in computational load and overfitting due to excessively dense local point clouds, or feature loss due to excessively sparse local point clouds. This ensures consistent sensitivity and reliability for all subsequent geometric analysis algorithms (e.g., fitting, curvature calculation) across different regions, improving the robustness and computational efficiency of the overall process. Step S21 specifically includes the following steps:

[0076] Step S211. Determine the voxel size based on the geometry of the anode guide rod. The voxel size is much smaller than the cross-sectional feature size of the guide rod, which is sufficient to preserve the subtle undulations of its edges and surfaces; at the same time, it can significantly reduce the amount of data in the original high-density point cloud, thus significantly improving the processing speed of subsequent algorithms. In this embodiment, the voxel size is set to 0.1 mm.

[0077] Step S212. Based on the boundaries and voxel dimensions of the 3D point cloud data, determine the number of voxels in each axis and construct a voxel network.

[0078] Step S213. Calculate the arithmetic mean of the three-dimensional coordinates for the point data within each voxel, and use the calculated value as the representative point of that voxel.

[0079] Step S22. Interpolation points are generated using the neighborhood point cloud interpolation method to fill in missing points in the 3D point cloud data. For local point cloud gaps on the guide rod surface caused by attachments, occlusion, or scanning blind spots, interpolation points are intelligently generated to fill these gaps by analyzing the spatial distribution patterns of effective points around the missing area (taking the average of neighborhood coordinates). This effectively repairs data "holes," avoids subsequent axis fitting distortion or cross-section center calculation errors caused by incomplete data, and significantly enhances the practicality and accuracy of the method under non-ideal scanning conditions.

[0080] Step S23. If the cross-section of the anode guide rod end face is a standard rectangular cross-section, then based on the point cloud data of the rectangular cross-section, the coordinates of the four vertices of the rectangular cross-section (x1,y1,0), (x2,y1,0), (x2,y2,0) and (x1,y2,0) are determined by the least squares matrix fitting method. The center coordinates are determined as the origin of the coordinate system by the coordinates of the four vertices. A unified measurement coordinate system is established with the extension direction of the anode guide rod as the Z-axis, the long side direction of the rectangular cross-section as the X-axis, and the short side direction of the rectangular cross-section as the Z-axis.

[0081] In this embodiment, based on the point cloud data of the anode guide rod end face, the least squares rectangle fitting method is used, and the Levenberg-Marquardt algorithm is used for iterative optimization to minimize the sum of squared algebraic distances from all points to the four sides of the rectangle, thus solving for the optimal rectangle parameters. The rectangle parameters include the center... Half-length Half width and the rotation angle around the direction of extension of the anode guide rod The coordinates of the four vertices (x1, y1, 0), (x2, y1, 0), (x2, y2, 0), and (x1, y2, 0) of the rectangular cross-section are determined using the rectangular parameters. The specific calculation method is as follows:

[0082] .

[0083] Step S24. If there is deformation in the cross-section of the anode guide rod end face, a segment of point cloud data is extracted along the extension direction of the anode guide rod. All point cloud data within this segment is projected and compressed onto the cross-section of the anode guide rod end face. The coordinates of the four vertices of the rectangular cross-section (x1, y1, 0), (x2, y1, 0), (x2, y2, 0), and (x1, y2, 0) are determined by fitting the data points after projection compression. The coordinates of the center are then determined as the origin of the coordinate system using the coordinates of the four vertices. A unified measurement coordinate system is established with the extension direction of the anode guide rod as the Z-axis, the long side of the rectangular cross-section as the X-axis, and the short side of the rectangular cross-section as the Y-axis. In this embodiment, a 20mm segment of point cloud data is extracted along the extension direction of the anode guide rod. It should be noted that the coordinates of the four vertices of the rectangular cross-section (x1, y1, 0), (x2, y1, 0), (x2, y2, 0), and (x1, y2, 0) are determined by fitting the data points after projection compression using the same iterative calculation method as in step S23.

[0084] Step S25. Calculate the single-point deviation of the 3D point cloud data. Based on a preset noise threshold For each data point in the 3D point cloud data, a judgment is made; if a single point deviates... Greater than the noise threshold If the data point is noisy, it is determined to be a noise point and deleted; otherwise, it is retained. This is based on a preset noise threshold. It effectively distinguishes real surface points from random noise points caused by scanning errors, flying points, or tiny floating objects. On the one hand, it has a high noise removal rate; on the other hand, it can preserve the real geometric details of the guide rod surface to the greatest extent, providing a "clean" data foundation for subsequent calculations and significantly reducing the impact of noise on the accuracy of axis fitting and quantitative analysis.

[0085] In this embodiment, the single-point deviation of 3D point cloud data The specific calculation method is as follows:

[0086] ,

[0087] in, Let be the X-axis coordinate value of the i-th point cloud data. This represents the mean of the point cloud data corresponding to the coordinate axes. Preset noise threshold. The specific calculation method is as follows:

[0088] ,

[0089] in, The standard deviation of the point cloud data corresponding to the coordinate axes. is the confidence coefficient. It should be noted that this embodiment uses the X-axis coordinate value of the point cloud data to remove noise points; however, the Y-axis or Z-axis coordinate values ​​can also be used, which will not be elaborated here.

[0090] Step S26. Based on the geometric dimensions of the anode guide rod, the background point data is deleted to obtain the main body point cloud data of the guide rod using an axisymmetric bounding box algorithm. Utilizing the known prior geometric dimensions of the anode guide rod (such as length, width, and height), an axisymmetric bounding box is defined to tightly enclose the theoretical space of the anode guide rod. All points outside the box (such as tooling table, ground, other equipment, and other background points) are automatically deleted at once. This method is highly efficient, processing a large number of point clouds at once. Furthermore, based on the design dimensions, it accurately separates the main body of the guide rod from the background, avoiding over-segmentation (erroneous deletion of guide rod points) or under-segmentation (residual background points). Step S26 specifically includes the following steps:

[0091] Step S261. Based on the geometric dimensions of the anode guide rod, set the boundary parameters of the axisymmetric bounding box. In this embodiment, the geometric dimensions of the anode guide rod are L×W. x ×W y Then the boundary of the axisymmetric bounding box is set as follows:

[0092] .

[0093] Step S262. Traverse each point in the 3D point cloud data. Based on the boundary of the axisymmetric bounding box, determine the state of each point. If the point is within the axisymmetric bounding box, it is determined that the point belongs to the main body of the guide rod and is retained. Otherwise, it is determined that the point is a background point or a noise point and is deleted.

[0094] Step S3. Based on the point cloud data of the guide rod body, construct the covariance matrix and solve for the eigenvalues ​​and corresponding eigenvectors. Use the eigenvector corresponding to the largest eigenvalue as the coarsely extracted central axis direction vector. Automatically calculate the main direction of the guide rod's overall extension, eliminating the need for manual experience-based judgment. This ensures the results are objective and repeatable, providing a core, traceable directional benchmark for all subsequent quantitative calculations. This is a crucial step in achieving automation and quantitative evaluation. Step S3 specifically includes the following steps:

[0095] Step S31. Based on the point cloud data of the guide rod body, construct the covariance matrix, specifically using the following calculation method:

[0096] ,

[0097] The covariance matrix C satisfies symmetry. , , , This indicates the correlation between the X-coordinate set and the Y-coordinate set. This indicates the correlation between the X-coordinate set and the Z-coordinate set. This indicates the correlation between the Y-coordinate set and the Z-coordinate set. These represent the autocorrelation of the corresponding coordinate sets. The elements of the covariance matrix C are specifically calculated using the following method:

[0098] ,

[0099] in, Let i be the three-dimensional coordinates of the i-th point cloud data. The mean of the point cloud data in the X, Y, and Z axes. This represents the total number of points in the point cloud.

[0100] Step S32. Based on the covariance matrix, solve for the eigenvalues ​​and corresponding eigenvectors, specifically using the following calculation method:

[0101] ,

[0102] in, It is a 3×3 identity matrix. For eigenvalues;

[0103] The solution obtained It includes three eigenvalues. and will The corresponding feature vector is the coarse extraction center axis direction vector. .

[0104] Step S4. Based on the coarsely extracted central axis direction vector, a circumscribed cuboid is obtained by fitting using the minimum volume method, and the central axis of the circumscribed cuboid is extracted to obtain the coarsely extracted central axis. For example... Figure 4 As shown, a tightly bounding cuboid is quickly fitted to the point cloud using the "minimum volume method". Its central axis stably reflects the overall spatial position and orientation of the guide rod, is insensitive to local surface defects, and exhibits strong robustness. It provides an excellent global initial value for subsequent refined piecewise fitting, preventing the iterative algorithm from getting stuck in local optima or diverging due to poor initial values, thus ensuring the stability and efficiency of the entire process. Step S4 specifically includes the following steps:

[0105] Step S41. Coarsely extract the direction vector of the central axis. The direction is Axis, construct a rectangular local coordinate system Based on the Cartesian local coordinate system and measurement coordinate system The coordinate rotation matrix R0 is determined. In this embodiment, the X-axis direction of the measurement coordinate system is selected as... The axis direction is determined by the right-hand rule. Axial direction.

[0106] Step S42. Transform the point cloud data of the guide rod body to the local coordinate system using the coordinate rotation matrix R0. .

[0107] Step S43. Based on the transformed guide rod body point cloud data, determine the extreme boundary of the guide rod body point cloud data in the local coordinate system.

[0108] Step S44. Based on the extreme boundary, determine the parameters of the circumscribed cuboid, specifically using the following calculation method:

[0109] ,

[0110] in, The coordinates of the center of the circumscribed cuboid. These are the length, width, and height of the circumscribed cuboid, respectively. Point cloud data of the guide rod body in The extreme boundary of the axis, Point cloud data of the guide rod body in The extreme boundary of the axis, The point cloud data of the guide rod body are respectively in The extreme boundary of the axis.

[0111] Step S45. Based on the parameters of the circumscribed cuboid, determine the central axis of the circumscribed cuboid, and obtain the coarsely extracted central axis by inverse transformation back to the measurement coordinate system using the coordinate rotation matrix R0. (Central axis of the circumscribed cuboid) The specific calculation method is as follows:

[0112] ,

[0113] Coarse extraction of the central axis The specific calculation method is as follows:

[0114] .

[0115] Step S5. Based on the coarse extraction of the central axis direction, the point cloud data of the guide rod body is divided into several segments. Based on the point cloud data of each segment of the guide rod body, the central axis of each segment is obtained through a segmented fitting method. By using a segmented strategy, the global problem is decomposed into multiple local problems, and each segment of the point cloud is fitted independently. The resulting local axis can accurately reflect the subtle bending and offset of that segment of the guide rod, avoiding the limitation of existing technologies that can only measure individual sections. It realizes continuous and dense geometric feature sampling along the entire length of the guide rod, providing a direct and refined data source for calculating the "overall bending degree" and "series of local eccentric displacements". Step S5 specifically includes the following steps:

[0116] Step S51. Divide the anode guide rod into N segments along the coarse extraction center axis to obtain the point cloud data of each segment. ,in Indicates the first The number of point cloud data for the main body of the guide rod segment.

[0117] Step S52. Construct a fitting error objective function based on the point cloud data of each guide rod segment, minimize the fitting error objective function, and obtain the local reference point and local direction vector for each segment. The specific calculation method is as follows:

[0118] ,

[0119] in, The local reference point for the k-th segment is, i.e. At the starting point of the central axis, This is the local direction vector for the Kth segment. In this embodiment, it is achieved by adjusting the objective function. Regarding the six parameters respectively By taking the partial derivatives and setting them equal to zero, we can obtain the system of equations for solving the optimal parameters. The specific calculation method is as follows:

[0120] .

[0121] Step S52. Based on each local reference point and local direction vector, obtain the central axis of each segment. The specific calculation method is as follows:

[0122] .

[0123] Step S6. Based on the coarsely extracted center axis, globally align each segment of the center axis to obtain the center axis of the anode guide rod. For each segment of the fitted local axis, which may have slight misalignments, perform global rotation and translation optimization based on the coarsely extracted axis to eliminate accumulated errors between segments, ensuring that the final spliced ​​full-length center axis is a smooth, continuous, and high-precision spatial curve. The finely extracted center axis is the gold standard for calculating the "overall average curvature" and any "local eccentric displacement of a cross-section," providing a direct, reliable, and authoritative final basis for subsequent threshold-based quantitative assessment of structural integrity (qualified / unqualified). Step S6 specifically includes the following steps:

[0124] Step S61. Sample the data points on each segment of the central axis to obtain the source point set. In this embodiment, 100 evenly distributed sampling points are taken on each segment of the central axis.

[0125] Step S62. Sample the data points on the coarse extraction center axis to obtain the target point set corresponding to the source point set. Among them, the source point set sampling points in With the target point set The segments with the same index and relative position Correspondingly.

[0126] Step S63. Based on the source point set With the target point set The alignment error objective function is constructed and implemented using the following calculation method:

[0127] ,

[0128] in, For the source set Data points in the middle, For the target point set Zhongyu The corresponding data points For the source set The amount of data in the middle For rotation matrix, It is a translation matrix.

[0129] Step S64. Iteratively solve the alignment error objective function until the iteration termination condition is met, and obtain the rotation matrix. Translation matrix Based on rotation matrix Translation matrix The center axis of each segment is corrected to obtain the center axis of the anode guide rod. Step S64 specifically includes the following steps:

[0130] Step S641. Establish the current source point set Each point in In the target point set Find the point with the closest Euclidean distance in the interval and use it as the new corresponding point. The specific calculation method is as follows:

[0131] ,

[0132] For the updated source set The target point set is calculated using Euclidean distance. The nearest point in the middle is taken as the new corresponding point. Here, the source point set... .

[0133] Step S642. Based on and The correspondence is determined iteratively, and the objective function to minimize the alignment error is solved using the following calculation method:

[0134] ,

[0135] In this embodiment, the solution is obtained using the Singular Value Decomposition (SVD) method, specifically including:

[0136] (1) Calculate the source point set and target point set The centroid is calculated using the following method:

[0137] ,

[0138] in, For the updated source point set The amount of data in the middle For the updated source point set Data points, For the target point set Zhongyu Corresponding points, For the updated source point set The center of mass, For the target point set The center of mass.

[0139] (2) Calculate the centroid coordinates, specifically using the following calculation method:

[0140] ,

[0141] in, for Decentrifugation The coordinates after, for Decentrifugation The coordinates after.

[0142] (3) Calculate the covariance matrix The specific calculation method is as follows:

[0143] .

[0144] (4) For the covariance matrix Perform singular value decomposition and obtain the rotation matrix for the current iteration step from the decomposition results. and parallel matrix The specific calculation method is as follows:

[0145] ,

[0146] in, and It is a 3×3 orthogonal matrix. It is a 3×3 diagonal matrix.

[0147] (5) Based on rotation matrix and parallel matrix Update the current source set The specific calculation method is as follows:

[0148] ,

[0149] in, Let j be the source point set updated after the j-th iteration. For the current source point set, cumulatively update the global transformation. and The specific calculation method is as follows:

[0150] .

[0151] (6) Based on the alignment error objective function, through the rotation matrix and parallel matrix Calculate alignment error ,like Less than the preset error threshold If the maximum number of iterations is reached, the iteration stops; otherwise, the above steps are repeated for the (j+1)th iteration. In this embodiment, a preset error threshold is used. Set it to 0.1.

[0152] Step S643. Based on the final global transformation and The center axis of each segment is corrected to obtain the center axis of the anode guide rod. The specific calculation method is as follows:

[0153] ,

[0154] in, The base point of the central axis of segment K. Let be the direction vector of the central axis of the k-th segment. The base point of the center axis of the corrected Kth segment, This is the direction vector of the central axis of the Kth segment after correction. For example... Figure 5 As shown, after global alignment This is the central axis of the anode guide rod.

[0155] Step S7. Calculate the local eccentric displacement based on the anode guide rod central axis and the guide rod body point cloud data. Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis .

[0156] Local eccentric displacement The specific calculation method is as follows:

[0157] (1) Define the analysis section along the central axis of the anode guide rod at fixed intervals. Define a series of analysis section locations Extract the analysis section Point cloud data. In this embodiment, the interval... Set it to 50.

[0158] (2) Based on the analysis section The point cloud data was used to fit a rectangle using the least squares method to obtain the geometric center of the fitted rectangle. It should be noted that the rectangle fitting method is implemented using the same iterative calculation method as step S23.

[0159] (3) at fixed intervals Extract the coordinates of the corresponding point on the central axis of the anode guide rod. Computational geometric center Coordinates of the corresponding point The Euclidean distance is calculated using the following method:

[0160] .

[0161] (4) The maximum value is taken as the local eccentric displacement. .

[0162] Overall curvature The specific calculation method is as follows:

[0163] (1) Along the central axis of the anode guide rod, at fixed intervals Extraction point set In this embodiment, the interval Set to 1.

[0164] (2) Based on point set Calculate the curvature set The specific calculation method is as follows:

[0165] .

[0166] (3) Based on curvature set Obtain the full-length curvature The specific calculation method is as follows:

[0167] ,

[0168] in, The curvature is the total length. This is the length of the central axis of the anode guide rod.

[0169] Distance between the central axis of the anode guide rod and the theoretical axis The specific calculation method is as follows:

[0170] (1) The theoretical axis is the design center axis of the anode guide rod. In the unified measurement coordinate system, it is defined as passing through the origin (0, 0, 0) and in the direction of the anode guide rod extension. The equation of the theoretical axis is as follows:

[0171] ,

[0172] in, This is the length of the central axis of the anode guide rod.

[0173] (2) Based on point set Based on the theoretical axis equation, the distance between each point and the theoretical axis is obtained, specifically using the following calculation method:

[0174] ,

[0175] in, For point Distance from the theoretical axis, For point The X-axis and Y-axis coordinates.

[0176] (3) Take the maximum value of the distances between the point and the theoretical axis as the distance between the center axis of the anode guide rod and the theoretical axis. .

[0177] Step S8. Based on a preset threshold, through local eccentric displacement Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis The calculation results are used to assess the structural integrity of the anode guide rod. On the one hand, there is a local eccentric displacement. It provides a comprehensive and objective data basis for assessing local deformation, including full-length curvature. This directly reflects the degree of "non-straightness" of the guide rod as a whole, and the distance between the central axis of the anode guide rod and the theoretical axis. This ensures that the baseline error of the entire process is controllable, thereby making the eccentric displacement calculated based on this axis... and overall curvature The results are reliable, providing a solid foundation for the final evaluation conclusion. On the other hand, by setting industrial thresholds, the above calculation results are automatically compared with the standards to establish an objective, unified, and executable quantitative evaluation standard, thereby achieving automatic decision-making.

[0178] The preset thresholds include the first threshold. Second threshold and the third threshold In this embodiment, the first threshold Set to 0.02, the second threshold. Set to 3, the third threshold Set to 0.8. If the distance between the center axis of the anode guide rod and the theoretical axis... Less than or equal to the first threshold Local eccentric displacement Less than or equal to the second threshold and overall curvature Less than or equal to the third threshold If the structure is intact, the anode guide rod is considered complete; otherwise, the anode guide rod is considered incomplete.

[0179] Example 2

[0180] This embodiment provides a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions that can be executed by the processor. When the computer device is running, the processor and the memory communicate through the bus. When the machine-readable instructions are executed by the processor, the steps of the anode guide rod center axis extraction method based on 3D point cloud as described in any of the preceding embodiments are performed.

[0181] This embodiment also provides a computer-readable storage medium storing a computer program, which, when run by a processor, executes the steps of the method for extracting the central axis of the anode guide rod based on 3D point cloud as described in any of the preceding embodiments.

[0182] In summary, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for extracting the central axis of an anode guide rod based on 3D point clouds, characterized in that, include: Collect 3D point cloud data of the anode guide rod; The 3D point cloud data is preprocessed to remove noise points and background points to obtain the main body point cloud data of the guide rod. Specifically, based on the geometric dimension data of the anode guide rod, the main body point cloud data of the guide rod is obtained by deleting background point data through an axisymmetric bounding box algorithm. Based on the point cloud data of the guide rod body, a covariance matrix is ​​constructed and the eigenvalues ​​and corresponding eigenvectors are obtained. The eigenvector corresponding to the largest eigenvalue is used as the coarse extraction of the central axis direction vector. Based on the coarse extraction of the central axis direction vector, the circumscribed cuboid is obtained by fitting using the minimum volume method, and the central axis of the circumscribed cuboid is extracted to obtain the coarse extraction of the central axis. Based on the coarse extraction of the central axis direction, the point cloud data of the guide rod body is divided into several segments. Based on each segment of the guide rod body point cloud data, the central axis of each segment is obtained through a segmented fitting method. Based on the coarsely extracted center axis, each segment of the center axis is globally aligned to obtain the center axis of the anode guide rod. The global alignment includes sampling data points on each segment of the center axis to obtain a source point set. Sample the data points on the coarse extraction central axis to obtain the target point set corresponding to the source point set. Based on the source point set With the target point set Construct an alignment error objective function, iteratively solve the alignment error objective function until the iteration termination condition is met, and obtain the rotation matrix. Translation matrix Based on the rotation matrix Translation matrix The center axis of each segment is corrected to obtain the center axis of the anode guide rod.

2. The method for extracting the central axis of an anode guide rod based on 3D point clouds according to claim 1, characterized in that, Also includes: Based on the central axis of the anode guide rod and the point cloud data of the guide rod body, the local eccentric displacement is calculated. Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis ; Based on a preset threshold, through the local eccentric displacement Overall curvature and the distance between the central axis of the anode guide rod and the theoretical axis The calculation results are used to assess the structural integrity of the anode guide rod.

3. The method for extracting the central axis of the anode guide rod based on 3D point cloud as described in claim 2, characterized in that: The preset threshold includes a first threshold. Second threshold and the third threshold ; If the distance between the central axis of the anode guide rod and the theoretical axis Less than or equal to the first threshold The aforementioned local eccentric displacement Less than or equal to the second threshold and the overall curvature Less than or equal to the third threshold If the structure is intact, the anode guide rod is determined to be complete; otherwise, the anode guide rod is determined to be incomplete.

4. The method for extracting the central axis of an anode guide rod based on 3D point clouds according to any one of claims 1-3, characterized in that, The preprocessing of the 3D point cloud data to remove noise and background points specifically includes the following steps: Calculate the single-point deviation of the 3D point cloud data Based on a preset noise threshold For each data point in the 3D point cloud data, a determination is made; if the single-point deviation... Greater than the noise threshold If the condition is met, the data point is determined to be a noise point and deleted; otherwise, the data point is retained.

5. The method for extracting the central axis of an anode guide rod based on 3D point clouds according to claim 4, characterized in that, The preprocessing of the 3D point cloud data to remove noise and background points specifically includes the following steps: The density of the 3D point cloud data is uniformized by using a voxel downsampling method. Interpolation points are generated using a domain point cloud interpolation method to fill in the missing points in the 3D point cloud data; If the cross-section of the anode guide rod end face is a standard rectangular cross-section, then based on the point cloud data of the rectangular cross-section, the coordinates of the four vertices of the rectangular cross-section (x1,y1,0), (x2,y1,0), (x2,y2,0) and (x1,y2,0) are determined by the least squares matrix fitting method. The center coordinates are determined as the origin of the coordinate system using the coordinates of the four vertices. A unified measurement coordinate system is established with the extension direction of the anode guide rod as the Z-axis, the long side direction of the rectangular cross-section as the X-axis, and the short side direction of the rectangular cross-section as the Z-axis. If the cross-section of the anode guide rod end face is deformed, a segment of point cloud data is extracted along the extension direction of the anode guide rod. All point cloud data in this segment are projected and compressed onto the cross-section of the anode guide rod end face. The coordinates of the four vertices of the rectangular cross-section (x1,y1,0), (x2,y1,0), (x2,y2,0), and (x1,y2,0) are determined by fitting the data points after projection and compression. The center coordinates are determined as the origin of the coordinate system using the coordinates of the four vertices. A unified measurement coordinate system is established with the extension direction of the anode guide rod as the Z-axis, the long side direction of the rectangular cross-section as the X-axis, and the short side direction of the rectangular cross-section as the Y-axis.

6. The method for extracting the central axis of an anode guide rod based on 3D point clouds according to any one of claims 1-3 or 5, characterized in that, Based on the point cloud data of the guide rod body, a covariance matrix is ​​constructed and the eigenvalues ​​and corresponding eigenvectors are obtained. The eigenvector corresponding to the largest eigenvalue is used as the coarse extraction center axis direction vector. The specific steps include the following: Based on the point cloud data of the guide rod body, the covariance matrix is ​​constructed, specifically using the following calculation method: ， The covariance matrix C satisfies symmetry. , , , This indicates the correlation between the X-coordinate set and the Y-coordinate set. This indicates the correlation between the X-coordinate set and the Z-coordinate set. This indicates the correlation between the Y-coordinate set and the Z-coordinate set. These represent the autocorrelation of the corresponding coordinate sets; Based on the covariance matrix, the eigenvalues ​​and corresponding eigenvectors are solved using the following calculation method: ， in, It is a 3×3 identity matrix; The solution obtained It includes three eigenvalues. and the The corresponding feature vector is the coarse extraction center axis direction vector. .

7. The method for extracting the central axis of an anode guide rod based on 3D point clouds according to any one of claims 1-3 or 5, characterized in that, Based on the coarsely extracted central axis direction, the point cloud data of the guide rod body is divided into several segments. Based on each segment of the guide rod body point cloud data, the central axis of each segment is obtained through a segmented fitting method, specifically including the following steps: The anode guide rod is divided into N segments along the coarse extraction central axis, and the point cloud data of each segment is obtained. ,in Indicates the first The number of point cloud data for the main body of the guide rod; Based on the point cloud data of each guide rod segment, a fitting error objective function is constructed. By minimizing the fitting error objective function, the local reference point and local direction vector of each segment are obtained. The specific calculation method is as follows: ， in, This is the local reference point for segment K, i.e. At the starting point of the central axis, This is the local direction vector of the Kth segment; Based on each local reference point and local direction vector, the central axis of each segment is obtained, specifically using the following calculation method: 。 8. The method for extracting the central axis of an anode guide rod based on 3D point clouds according to any one of claims 1-3 or 5, characterized in that, The process of globally aligning each segment of the center axis based on the coarsely extracted center axis to obtain the anode guide rod center axis also includes: Based on the source point set With the target point set The alignment error objective function is constructed and implemented using the following calculation method: ， in, For the source set Data points in the middle, For the target point set Zhongyu The corresponding data points For the source set The amount of data in the middle For rotation matrix, It is a translation matrix.

9. A computer device, characterized in that, include: The system includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the method for extracting the central axis of an anode guide rod based on 3D point clouds as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the method for extracting the central axis of an anode guide rod based on 3D point clouds as described in any one of claims 1-8.

Images