A method for measuring the structure height of a hybrid insulator parison positioning surface

Abstract

The application belongs to the technical field of insulator size detection, and relates to a mixed insulator parison positioning surface structure height measuring method. A spring clamp with a reference plate bonded is clamped at the insulator parison umbrella skirt position; top view point cloud and bottom view point cloud are respectively acquired by scanning in a normal posture and an inverted posture; top view reference plate point cloud, bottom view reference plate point cloud, upper positioning surface point cloud and lower positioning surface point cloud are extracted; an initial alignment matrix is calculated through the centroid; one-stage registration is performed on the reference plate edge points to obtain a one-stage transformation matrix; after the bottom view reference plate point cloud is applied with the one-stage transformation matrix, two-stage registration is performed to obtain a two-stage transformation matrix; after translation, a target transformation matrix is obtained to align the upper and lower positioning surfaces, and then the average height difference is calculated to obtain the positioning surface structure height. Without the need of pasting mark points, the two-stage registration effectively suppresses the slip phenomenon in point cloud registration, and the method has high measurement precision and high efficiency, and is suitable for batch measurement of the positioning surface structure height of the insulator parison.

Description

Technical Field

[0001] This invention relates to the field of insulator size detection technology, and in particular to a method for measuring the height of the positioning surface structure of a hybrid insulator blank. Background Technology

[0002] As a critical component in power transmission lines, the quality of power insulators directly impacts the safety of power grid operation. In recent years, a new type of hybrid insulator has been developed. Using traditional porcelain or glass insulators as preforms, it is manufactured through a high-temperature injection molding process, forming a solid high-temperature vulcanized silicone rubber (HTV) coating layer in one step. This design aims to improve anti-flashover performance and reduce manual maintenance costs. In the manufacturing process of hybrid insulators, the position used to pressurize and fix the insulator preform is called the positioning surface, which includes an upper positioning surface and a lower positioning surface. The height distance between the upper and lower positioning surfaces is the positioning surface structural height. The quality of the positioning surface structural height significantly affects the molding process of the hybrid insulator. Therefore, before molding, it is necessary to screen the insulator preforms to be processed, discarding unqualified preforms to ensure the processing quality and stable production of hybrid insulators.

[0003] Because the insulator blank is a hollow cylindrical structure, with the upper and lower positioning surfaces located on its inner and outer sides respectively, conventional contact measuring tools (such as calipers and depth gauges) cannot simultaneously penetrate the inner cavity and cross the cylinder wall to conform to both reference surfaces, making it difficult to accurately measure the height of the positioning surface structure. Non-contact measurements, on the other hand, cannot simultaneously acquire the spatial positions of the two positioning surfaces from a single viewpoint, thus also making it difficult to measure the height of the positioning surface structure. If the height of the positioning surface structure is calculated by extracting the point clouds of the upper and lower positioning surfaces from the complete three-dimensional point cloud data of the insulator blank, it typically requires using reflective markers combined with laser scanning and Iterative Closest Point (ICP) point cloud registration, followed by point cloud post-processing. This process is cumbersome, inefficient, and fails to meet the batch testing requirements of actual production. Furthermore, the adhesive used to paste the markers easily leaves residue on the surface of the insulator blank, causing contamination.

[0004] In summary, the existing methods for measuring the height of the positioning surface structure of insulator blanks suffer from problems such as inaccurate measurement results, cumbersome measurement process, low efficiency, and contamination of the insulator blanks, making it difficult to meet the batch testing requirements in actual production. Summary of the Invention

[0005] Therefore, the technical problem to be solved by the present invention is to overcome the problems of inaccurate measurement results, cumbersome measurement process, low efficiency, and pollution of insulator blanks in the existing measurement method for the positioning surface structure height of insulator blanks, which makes it difficult to meet the batch testing requirements in actual production.

[0006] To solve the above-mentioned technical problems, the present invention provides a method for measuring the height of the positioning surface structure of a hybrid insulator blank, comprising: A reference plate is set at the shed part of the insulator blank so that a fixed relative positional relationship is formed between the reference plate and the insulator blank; After the insulator blank is placed upright on the horizontal plane, it is scanned to obtain the top view point cloud data, thereby extracting the top view reference plate point cloud and the upper positioning surface point cloud; after the insulator blank is placed upside down on the horizontal plane, it is scanned to obtain the bottom view point cloud data, thereby extracting the bottom view reference plate point cloud and the lower positioning surface point cloud. Calculate the first centroid of the top-view reference plate point cloud and the second centroid of the bottom-view reference plate point cloud, and construct the initial alignment transformation matrix of the top-view point cloud and the bottom-view point cloud based on the first centroid and the second centroid; Based on the point clouds of the bottom and top reference plates, source point sets and target point sets are obtained. The objective function is to minimize the positional deviation between the source point set and the target point set after rigid rotation transformation and translation compensation. The stage transformation matrix is ​​obtained by iteratively solving the objective function. The target bottom reference plate point cloud is obtained by transforming the bottom reference plate point cloud using the stage transformation matrix. The thickness of the reference plate is used as the translation amplitude and the normal vector direction of the target bottom reference plate point cloud is used as the translation direction to construct the translation matrix. The target transformation matrix is ​​constructed based on the initial alignment transformation matrix, the stage transformation matrix, and the translation matrix; the target transformation matrix is ​​used to transform the lower positioning surface point cloud to obtain the target lower positioning surface point cloud; the positioning surface structure height of the insulator blank is obtained based on the upper positioning surface point cloud and the target lower positioning surface point cloud.

[0007] Preferably, the first centroid of the top-view reference plate point cloud and the second centroid of the bottom-view reference plate point cloud are calculated, and an initial alignment transformation matrix for the top-view and bottom-view point clouds is constructed based on the first and second centroids, including: The three-dimensional coordinates of the first centroid are obtained based on the average of the three-dimensional coordinates of each point in the top-view reference plate point cloud; the three-dimensional coordinates of the second centroid are obtained based on the average of the three-dimensional coordinates of each point in the bottom-view reference plate point cloud. Based on the x-axis and y-axis coordinates of the first centroid and the x-axis and y-axis coordinates of the second centroid, calculate the rotation angle of the second centroid to the first centroid about the z-axis; Based on the z-axis coordinates of the first centroid and the second centroid, calculate the translation distance along the z-axis from the second centroid to the first centroid. The initial alignment transformation matrix for the top and bottom viewpoint clouds is constructed based on the rotation angle and translation distance.

[0008] Preferably, the three-dimensional coordinates of the first centroid Represented as: , in, The first point cloud in the top-view reference board represents the first... The three-dimensional coordinates of each point; This represents the total number of points in the point cloud of the top-view reference board; The three-dimensional coordinates of the second centroid Represented as: , in, The first point cloud in the bottom reference plate represents the... The three-dimensional coordinates of each point; This represents the total number of points in the point cloud of the bottom-view reference plate; The rotation angle from the second mass center to the first mass center about the z-axis Represented as: , Translational distance along the z-axis from the second centroid to the first centroid Represented as: , Initial alignment transformation matrix Represented as: .

[0009] Preferably, source point sets and target point sets are obtained based on the point clouds of the bottom-view reference plate and the top-view reference plate. The objective is to minimize the positional deviation between the source point set and the target point set after rigid rotation transformation and translation compensation. An objective function is constructed, and the stage transformation matrix is ​​obtained by iteratively solving the objective function, including: Extract the first edge point set from the top-view reference plate point cloud and the second edge point set from the bottom-view reference plate point cloud; The first objective function is constructed by taking the second set of edge points as the source point cloud and the first set of edge points as the target point cloud. The objective is to minimize the sum of the distances between each target source point after coordinate transformation by the first rotation matrix and the first translation vector in the source point cloud and each target point in the target point cloud that is closest to each source point. Based on the values ​​of the translation along the x-axis, the translation along the y-axis, and the rotation angle around the z-axis before and after the coordinate transformation, the first boundary constraint condition is constructed. The L-BFGS-B algorithm is used to iteratively solve the first objective function based on the first boundary constraint condition until the difference between the values ​​of the first objective function after two iterations is less than a preset function threshold, or the projection gradient of the first rotation matrix and the first translation vector after the iteration is less than a preset gradient threshold, or the number of iterations reaches a preset number of iterations. The first rotation matrix and the first translation vector are then obtained, thus obtaining the first-stage transformation matrix.

[0010] Preferably, the step of extracting the edge point set of the reference plate point cloud includes: Each point in the point cloud of the baseline board is used as the point to be detected; A radius search is used to obtain the neighborhood point set of the point to be detected, and the azimuth angle of each neighbor point in the neighborhood point set relative to the point to be detected is calculated. The azimuth angles of all the neighbor points are arranged from largest to smallest to obtain an azimuth angle sequence. The difference between each azimuth angle in the azimuth angle sequence and the next azimuth angle is calculated. If the maximum value of all the calculated differences is greater than or equal to a preset angle threshold, the point to be detected is determined to be an edge point. If the maximum value of all the calculated differences is less than the preset angle threshold, the point to be detected is determined to be an interior point. Specifically, when calculating the difference between each azimuth angle in the azimuth angle sequence and the next azimuth angle, for the last azimuth angle, the difference between that azimuth angle and the first azimuth angle in the azimuth angle sequence is calculated. The sum of.

[0011] Preferably, the first objective function is expressed as: , in, Represents the first rotation matrix; Indicates the first translation vector; Indicates the second edge point set. One point; Indicates the first edge point set with The nearest point; Indicates the number of points in the second edge point cluster; Represents the Euclidean norm; The first boundary constraint condition is expressed as follows: , in, This represents the amount of translation along the x-axis before and after the coordinate transformation; This indicates the lower bound of the translation along the x-axis; This represents the upper bound of the translation along the x-axis; This represents the amount of translation along the y-axis before and after the coordinate transformation; This indicates the lower bound of the translation along the y-axis; This represents the upper bound of the translation along the y-axis; Represents the rotation angle around the z-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the z-axis; This represents the upper bound of the rotation angle about the z-axis.

[0012] Preferably, after obtaining the first-stage transformation matrix, the process further includes: A first-stage transformed bottom reference plate point cloud is obtained by performing coordinate transformation on the bottom reference plate point cloud using a one-stage transformation matrix. The bottom view reference plate point cloud is used as the source point cloud and the top view reference plate point cloud is used as the target point cloud. The objective function is to minimize the sum of the vertical distances between each target source point in the source point cloud after coordinate transformation by the second rotation matrix and the second translation vector and the tangent plane of each target point in the target point cloud that is closest to each source point. Based on the values ​​of the rotation angle around the x-axis, the rotation angle around the y-axis, and the translation along the z-axis before and after the coordinate transformation, a second boundary constraint condition is constructed. The L-BFGS-B algorithm is used to iteratively solve the second objective function based on the second boundary constraint condition until the difference between the values ​​of the second objective function after two iterations is less than a preset function threshold, or the projection gradient of the second rotation matrix and the second translation vector after the iteration is less than a preset gradient threshold, or the number of iterations reaches a preset number of iterations. The second rotation matrix and the second translation vector are obtained, thus obtaining the two-stage transformation matrix. The target bottom-view reference plate point cloud is obtained by transforming the bottom-view reference plate point cloud using the two-stage transformation matrix.

[0013] Preferably, the second objective function is expressed as: , in, Indicates the second rotation matrix; Indicates the second translation vector; This represents the first transformation of the bottom-view reference plate point cloud. One point; Indicates the point cloud of the top-view reference board and The nearest point; express The unit normal vector at that location; Represents the dot product operation of vectors; The second boundary constraint condition is expressed as follows: , in, Represents the rotation angle around the x-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the x-axis; This represents the upper bound of the rotation angle about the x-axis; Indicates the rotation angle around the y-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the y-axis; This represents the upper bound of the rotation angle about the y-axis; This represents the translation along the z-axis before and after the coordinate transformation; This represents the lower bound of the translation along the z-axis. This represents the upper bound of the translation along the z-axis.

[0014] Preferably, the target transformation matrix Represented as: , in, Represents the translation matrix; Represents the two-stage transformation matrix; Represents the first-stage transformation matrix; This represents the initial alignment transformation matrix.

[0015] Preferably, the positioning surface structure height of the insulator blank is obtained based on the point cloud of the upper positioning surface and the point cloud of the target lower positioning surface, including: The first average height of the upper positioning surface is obtained based on the average z-axis coordinate of each point in the point cloud of the upper positioning surface. The second average height of the lower positioning surface is obtained based on the average z-axis coordinate of each point in the point cloud of the target lower positioning surface. The positioning surface structure height of the insulator blank is obtained based on the difference between the first average height and the second average height.

[0016] The method for measuring the structural height of the positioning surface of the hybrid insulator blank provided in this application has the following advantages: 1. By setting a reference plate at the umbrella skirt of the insulator blank, the reference plate and the insulator blank maintain a fixed relative pose, providing a unified rigid reference for the two scans. After completing the top and bottom scans by placing the insulator blank in the forward and inverted positions respectively, the reference plate and the point clouds of the upper and lower positioning surfaces are separated and extracted. Based on the centroids of the top and bottom reference plate point clouds, an initial alignment transformation matrix is ​​constructed to achieve coarse registration of the upper and lower point clouds. Then, based on the bottom reference plate point clouds, the source point cloud and the target point cloud are obtained. A rigid registration objective function with minimum positional deviation is constructed and iteratively solved to obtain the stage transformation matrix to complete the high-precision fine registration of the top and bottom point clouds. Finally, the normal of the finely registered bottom reference plate point cloud is used as the translation direction and the thickness of the reference plate is used as the translation amplitude to generate a translation matrix to compensate for the thickness of the reference plate itself during forward and reverse scanning. The axial position offset is fused with the initial alignment transformation matrix, the stage transformation matrix, and the translation matrix to obtain the complete target transformation matrix. This target transformation matrix is ​​used to perform coordinate transformation on the lower positioning surface point cloud obtained by inverted scanning so that it is in the same coordinate system as the upper positioning surface point cloud obtained by forward scanning. Finally, the structural height of the positioning surface of the insulator blank is accurately calculated by combining the upper positioning surface point cloud and the corrected target lower positioning surface point cloud in the same coordinate system. This application uses a planar reference plate fixed and bonded on the spring clip as a spatial reference. Only the spring clip needs to be held at the umbel skirt position of the insulator blank to establish the registration reference between the top view point cloud and the bottom view point cloud. There is no need to paste a large number of marker points, and there is no need to rely on a turntable or multi-view splicing system. The measurement can be completed by using only a line laser scanner to obtain the top view point cloud and the bottom view point cloud. The operation process is simple, the measurement cost is low, and the measurement efficiency is high. 2. To address the issues of slippage and getting trapped in local optima in traditional ICP algorithms for point cloud registration of planar reference plates, this application employs the L-BFGS-B algorithm to design a two-stage registration strategy. The first stage uses the edge point sets of the bottom-view and top-view reference plate point clouds as the source and target point sets for registration. The second stage uses the bottom-view and top-view reference plate point clouds after a single registration transformation as the source and target point sets for registration. The two registration stages converge progressively, effectively suppressing slippage, improving the registration accuracy of the top-view and bottom-view reference plate point clouds, and reducing the measurement error of the insulator blank positioning surface structure height. Attached Figure Description

[0017] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein: Figure 1 Flowchart of the method for measuring the height of the positioning surface structure of the hybrid insulator blank provided in this application; Figure 2 A schematic diagram showing the positions of the upper and lower positioning surfaces of the insulator blank provided in this application; Figure 3 A schematic diagram of the spring clip holding the insulator blank skirt part provided in this application; Figure 4 This is a schematic diagram showing the extraction of the center point of the insulator blank provided in this application; Figure 5 A schematic diagram of the registration of edge points of the planar reference plate provided in this application; Figure 6 A schematic diagram of the point cloud of the planar reference plate before registration, the upper positioning surface, and the lower positioning surface provided for this application; wherein, Figure 6 (a) in the figure is a schematic diagram of the point cloud of the planar reference plate before registration. Figure 6 (b) in the diagram is a schematic diagram of the upper and lower positioning surfaces before registration; Figure 7 A schematic diagram of the registered planar reference plate point cloud, upper positioning surface, and lower positioning surface provided for this application; wherein, Figure 7 (a) in the figure is a schematic diagram of the point cloud of the registered planar reference plate. Figure 7 (b) in the diagram is a schematic diagram of the upper and lower positioning surfaces after registration; Explanation of the reference numerals in the accompanying drawings: 1. Insulator blank; 11. Upper positioning surface; 12. Lower positioning surface; 2. Spring clip; 3. Reference plate. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0019] Please see Figure 1 , Figure 1 The diagram shows a flowchart of the method for measuring the height of the positioning surface structure of the hybrid insulator blank provided in this application. The method includes: S10~S60.

[0020] S10: A reference plate is set at the shed part of the insulator blank so that a fixed relative positional relationship is formed between the reference plate and the insulator blank.

[0021] Optionally, after a reference plate is bonded to the outer side of one clamping surface of the spring clip, the spring clip is clamped to the shed portion of the insulator blank, thereby establishing a fixed relative positional relationship between the reference plate and the insulator blank. Specifically, as shown... Figure 2 and Figure 3 As shown, the upper positioning surface 11 of the insulator blank 1 is located on the outside and the lower positioning surface 12 is located on the inside. By attaching the reference plate 3 to the outside of one clamping surface of the spring clip 2 and clamping it on the skirt part of the insulator blank 1, a fixed relative positional relationship can be formed between the reference plate 3 and the insulator blank 1.

[0022] In a specific example, the reference plate 3 is a carbon fiber planar reference plate with a length of 70m, a width of 30m, and a thickness of 1.6mm. The reference plate 3 is glued to the spring clip 2 using 502 glue and then clamped at the position of the first shed from the top of the insulator blank 1 to be measured, so that the reference plate 3 and the insulator blank 1 maintain a fixed relative positional relationship.

[0023] S20: After placing the insulator blank upright on the horizontal plane, scan it to obtain the top view point cloud data and extract the top view reference plate point cloud and the upper positioning surface point cloud; after placing the insulator blank upside down on the horizontal plane, scan it to obtain the bottom view point cloud data and extract the bottom view reference plate point cloud and the lower positioning surface point cloud.

[0024] Specifically, in one embodiment of this application, the insulator blank is placed upright on a horizontal plane, and a lead screw motor drives a line laser scanner (model MV-DP3580-01P) to scan the insulator blank and the reference plate at a frame rate of 520Hz to obtain top viewpoint cloud data; then the insulator blank is rotated 180° and placed in an inverted position on a horizontal plane, and a lead screw motor drives a line laser scanner (model MV-DP3580-01P) to scan the insulator blank and the reference plate at a frame rate of 520Hz to obtain bottom viewpoint cloud data.

[0025] Furthermore, point cloud processing algorithms are used to extract the top view reference plate point cloud and the upper positioning surface point cloud from the top view point cloud data, and to extract the bottom view reference plate point cloud and the lower positioning surface point cloud from the bottom view point cloud data. Specifically, this includes steps 1-1 to 1-4.

[0026] Step 1-1: Set the height threshold The top-view and bottom-view point cloud data were filtered using a pass-through filtering method to remove point clouds with z-axis coordinates less than -25mm.

[0027] Steps 1-2: Project the filtered top-view and bottom-view point cloud data onto the XOY plane, transforming the coordinates of each point in the top-view and bottom-view point cloud data from (x, y, z) to (x, y, 0). Use a contour extraction algorithm to extract the maximum contour of the projected point cloud, and then perform circle fitting on the maximum contour using the least squares method. Use the center of the fitted circle as the center of the insulator blank. Figure 4 As shown.

[0028] Steps 1-3: In the top view point cloud data, extract a point cloud with a radius of 47.5mm centered on the center of the insulator blank as the upper positioning surface point cloud; in the bottom view point cloud data, extract a point cloud with a radius of 70mm centered on the center of the insulator blank as the lower positioning surface point cloud.

[0029] Steps 1-4: In the top-view point cloud data, using the center of the insulator blank as the center, a pass-through filter is used to exclude point clouds of the insulator blank body with a radius less than 200mm. For the remaining point clouds, a density-based spatial clustering of applications with noise (DBSCAN) algorithm is used for clustering, and the largest cluster is extracted as the top-view reference plate point cloud. In the bottom-view point cloud data, using the center of the insulator blank as the center, a pass-through filter is used to exclude point clouds of the insulator blank body with a radius less than 200mm. For the remaining point clouds, a DBSCAN algorithm is used for clustering, and the largest cluster is extracted as the bottom-view reference plate point cloud.

[0030] S30: Calculate the first centroid of the top-view reference plate point cloud and the second centroid of the bottom-view reference plate point cloud, and construct the initial alignment transformation matrix of the top-view point cloud and the bottom-view point cloud based on the first centroid and the second centroid.

[0031] Further, step S30 includes: S300~S303.

[0032] S300: Based on the average of the three-dimensional coordinates of each point in the top-view reference plate point cloud, the three-dimensional coordinates of the first centroid are obtained; based on the average of the three-dimensional coordinates of each point in the bottom-view reference plate point cloud, the three-dimensional coordinates of the second centroid are obtained.

[0033] Specifically, the three-dimensional coordinates of the first centroid Represented as: , in, The first point cloud in the top-view reference board represents the first... The three-dimensional coordinates of each point; This represents the total number of points in the point cloud of the top-view reference plate.

[0034] Specifically, the three-dimensional coordinates of the second centroid Represented as: , in, The first point cloud in the bottom reference plate represents the... The three-dimensional coordinates of each point; This represents the total number of points in the point cloud of the bottom reference plate.

[0035] S301: Based on the x-axis and y-axis coordinates of the first centroid and the x-axis and y-axis coordinates of the second centroid, calculate the rotation angle of the second centroid to the first centroid about the z-axis.

[0036] Specifically, the rotation angle from the second centroid to the first centroid about the z-axis Represented as: .

[0037] S302: Based on the z-axis coordinates of the first and second centroids, calculate the translation distance along the z-axis from the second centroid to the first centroid.

[0038] Specifically, the translational distance along the z-axis from the second centroid to the first centroid. Represented as: .

[0039] S303: Construct the initial alignment transformation matrix for the top and bottom viewpoint clouds based on the rotation angle and translation distance.

[0040] Specifically, the initial alignment transformation matrix Represented as: .

[0041] S40: Based on the point cloud of the bottom reference plate and the point cloud of the top reference plate, obtain the source point cloud and the target point cloud. With the goal of minimizing the positional deviation between the source point cloud and the target point cloud after rigid rotation transformation and translation compensation, construct the objective function and iteratively solve the objective function to obtain the stage transformation matrix.

[0042] Further, step S40 includes: S400~S407.

[0043] S400: Extract the first edge point set of the top-view reference plate point cloud and the second edge point set of the bottom-view reference plate point cloud.

[0044] Specifically, step S400 includes: steps 2-1 to 2-2.

[0045] Step 2-1: Take each point in the point cloud of the reference board as the point to be detected.

[0046] Step 2-1: Use radius search to obtain the neighborhood point set of the point to be detected, and calculate the azimuth angle of each neighborhood point relative to the point to be detected. Arrange the azimuth angles of all neighborhood points from largest to smallest to obtain the azimuth angle sequence. Calculate the difference between each azimuth angle in the azimuth angle sequence and the next azimuth angle. If the maximum value of all the calculated differences is greater than or equal to the preset angle threshold, the point to be detected is determined to be an edge point. If the maximum value of all the calculated differences is less than the preset angle threshold, the point to be detected is determined to be an interior point.

[0047] Specifically, when calculating the difference between each azimuth angle in the azimuth angle sequence and the next azimuth angle, for the last azimuth angle, the difference between that azimuth angle and the first azimuth angle in the azimuth angle sequence is calculated. The sum of.

[0048] For example, for a point P to be detected, searching for at least 5 neighboring points within a radius of 2.5 mm yields a set of neighboring points. Calculate the azimuth angle of each neighboring point relative to the point P to be detected: , in, This represents the azimuth angle of the i-th neighboring point relative to the point to be detected, P; Represents the x-axis and y-axis coordinates of the i-th neighboring point; This represents the x-axis and y-axis coordinates of the point P to be detected.

[0049] Calculate the difference between each azimuth angle in the azimuth angle sequence and the next azimuth angle: , in, This represents the difference between the i-th azimuth and the (i+1)-th azimuth in the azimuth sequence; This represents the difference between the last azimuth and the first azimuth in the azimuth sequence. The sum of.

[0050] Select the maximum difference: ,like ,in, If the preset angle threshold is used, the detection point P is determined to be an edge point; otherwise, the detection point P is an interior point.

[0051] S401: Using the second set of edge points as the source point cloud and the first set of edge points as the target point cloud, the objective function is constructed with the goal of minimizing the sum of the distances between each target source point in the source point cloud (after coordinate transformation by the first rotation matrix and the first translation vector) and each target point in the target point cloud that is closest to each source point.

[0052] Specifically, the first objective function is expressed as: , in, Represents the first rotation matrix; Indicates the first translation vector; Indicates the second edge point set. One point; Indicates the first edge point set with The nearest point; Indicates the number of points in the second edge point cluster; This represents the Euclidean norm.

[0053] S402: Based on the values ​​of the translation along the x-axis, the translation along the y-axis, and the rotation angle around the z-axis before and after the coordinate transformation, construct the first boundary constraint conditions.

[0054] Specifically, the first boundary constraint condition is expressed as: , in, This represents the amount of translation along the x-axis before and after the coordinate transformation; This indicates the lower bound of the translation along the x-axis; This represents the upper bound of the translation along the x-axis; This represents the amount of translation along the y-axis before and after the coordinate transformation; This indicates the lower bound of the translation along the y-axis. This represents the upper bound of the translation along the y-axis; Represents the rotation angle around the z-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the z-axis; This represents the upper bound of the rotation angle about the z-axis.

[0055] In a specific example, , , , , , .

[0056] S403: The Limited-memory-Broyden-Fletcher-Goldfarb-Shanno with Box constraints (L-BFGS-B) algorithm is used to iteratively solve the first objective function based on the first boundary constraint condition until the difference between the values ​​of the first objective function after two iterations is less than a preset function threshold, or the projected gradient of the first rotation matrix and the first translation vector after the iteration is less than a preset gradient threshold, or the number of iterations reaches a preset number of iterations. The first rotation matrix and the first translation vector are obtained, thus obtaining the first-stage transformation matrix.

[0057] For example, the preset function threshold can be 1×10. -10 The preset gradient threshold can be 1×10. -10 The preset number of iterations can be 500.

[0058] S404: After performing coordinate transformation on the bottom reference plate point cloud using a one-stage transformation matrix, a first-stage transformed bottom reference plate point cloud is obtained.

[0059] S405: Using the bottom view reference plate point cloud as the source point cloud and the top view reference plate point cloud as the target point cloud, the objective function is to minimize the sum of the vertical distances between each target source point in the source point cloud after coordinate transformation by the second rotation matrix and the second translation vector and the tangent plane of each target point in the target point cloud that is closest to each source point.

[0060] Specifically, the second objective function is expressed as: , in, Indicates the second rotation matrix; Indicates the second translation vector; This represents the first transformation of the bottom-view reference plate point cloud. One point; Indicates the point cloud of the top-view reference board and The nearest point; express The unit normal vector at that location; This represents the dot product operation of vectors.

[0061] S406: Based on the values ​​of the rotation angle around the x-axis, the rotation angle around the y-axis, and the translation along the z-axis before and after the coordinate transformation, construct the second boundary constraint conditions.

[0062] Specifically, the second boundary constraint condition is expressed as: , in, Represents the rotation angle around the x-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the x-axis; This represents the upper bound of the rotation angle about the x-axis; Indicates the rotation angle around the y-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the y-axis; This represents the upper bound of the rotation angle about the y-axis; This represents the translation along the z-axis before and after the coordinate transformation; This represents the lower bound of the translation along the z-axis. This represents the upper bound of the translation along the z-axis.

[0063] In a specific example, , , , , , .

[0064] S407: The L-BFGS-B algorithm is used to iteratively solve the second objective function based on the second boundary constraint condition until the difference between the values ​​of the second objective function after two iterations is less than the preset function threshold, or the projection gradient of the second rotation matrix and the second translation vector after the iteration is less than the preset gradient threshold, or the number of iterations reaches the preset number of iterations. The second rotation matrix and the second translation vector are obtained, thus obtaining the two-stage transformation matrix. The target bottom-view reference plate point cloud is obtained by transforming the bottom-view reference plate point cloud using the two-stage transformation matrix.

[0065] S50: After transforming the bottom reference plate point cloud using the stage transformation matrix, the target bottom reference plate point cloud is obtained. The thickness of the reference plate is used as the translation amplitude, and the normal vector direction of the target bottom reference plate point cloud is used as the translation direction to construct the translation matrix.

[0066] Specifically, in response to the axial coordinate difference caused by the thickness of the reference plate during forward and reverse scanning, this application constructs a translation matrix that is used only for thickness compensation, with the vertical direction of the registered reference plate surface as the moving direction and the plate thickness as the moving distance, in order to eliminate the system measurement error caused by the thickness of the reference plate.

[0067] S60: Construct the target transformation matrix based on the initial alignment transformation matrix, the stage transformation matrix, and the translation matrix; transform the lower positioning surface point cloud using the target transformation matrix to obtain the target lower positioning surface point cloud; obtain the positioning surface structure height of the insulator blank based on the upper positioning surface point cloud and the target lower positioning surface point cloud.

[0068] Specifically, the target transformation matrix Represented as: , in, Represents the translation matrix; Represents the two-stage transformation matrix; Represents the first-stage transformation matrix; This represents the initial alignment transformation matrix.

[0069] In a specific example, the initial alignment transformation for: , One-stage transformation matrix for: , Two-stage transformation matrix for: , Translation matrix for: , The target transformation matrix is: .

[0070] Furthermore, based on the point cloud of the upper positioning surface and the point cloud of the target lower positioning surface, the positioning surface structure height of the insulator blank is obtained, including: steps 3-1 to 3-3.

[0071] Step 3-1: Based on the average z-axis coordinates of each point in the upper positioning surface point cloud, obtain the first average height of the upper positioning surface.

[0072] Step 3-2: Based on the average z-axis coordinates of each point in the target lower positioning surface point cloud, obtain the second average height of the lower positioning surface.

[0073] Step 3-3: Obtain the positioning surface structure height of the insulator blank based on the difference between the first average height and the second average height.

[0074] Specifically, the height of the positioning surface structure The calculation formula is: , in, Indicates the first average height; This indicates the second average height.

[0075] In this embodiment, the above method is used to measure the height of the positioning surface structure of 10 300-KN type glass insulator blanks, such as... Figure 5 The diagram shown is a schematic diagram of edge point registration of a planar reference plate provided in an embodiment of this application. Figure 6 The diagram shown is a schematic representation of the point cloud, upper positioning surface, and lower positioning surface of the planar reference plate before registration provided in this application. Figure 6 (a) in the figure is a schematic diagram of the point cloud of the planar reference plate before registration. Figure 6 (b) in the diagram is a schematic diagram of the upper and lower positioning surfaces before registration. Figure 7 The diagram shown is a schematic representation of the point cloud, upper positioning surface, and lower positioning surface of the registered planar reference plate provided in this application. Figure 7 (a) in the figure is a schematic diagram of the point cloud of the registered planar reference plate. Figure 7 (b) in the diagram shows the upper and lower positioning surfaces after registration. (By comparison) Figure 6 and Figure 7 It can be seen that after registration using the method provided in this application, the point clouds of the top and bottom view reference plates, which were originally misaligned and had obvious rotation and translation deviations, are closely aligned without obvious slippage or offset. At the same time, the point cloud of the lower positioning surface acquired inverted is unified with the point cloud of the upper positioning surface acquired in the forward direction to the same coordinate system after coordinate transformation. The spatial position correspondence between the upper and lower positioning surfaces is clear and there is no coordinate misalignment.

[0076] Table 1 shows a comparison of the measurement results of the embodiments of this application with the measurement results of the traditional ICP registration method: Table 1

[0077] As shown in Table 1, the average absolute error of the traditional ICP method is 0.25 mm, with a maximum error of 0.61 mm. In contrast, the method provided in this application (L-BFGS-B two-stage registration) has an average absolute error of only 0.13 mm and a maximum error of 0.28 mm. Compared to the ICP method, the method provided in this application reduces the average measurement error by 48% and the maximum error by 54%. These results demonstrate that this application effectively suppresses slippage during point cloud registration and significantly improves the accuracy and stability of the positioning surface structure height measurement.

[0078] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for measuring the height of the positioning surface structure of a hybrid insulator blank, characterized in that, include: A reference plate is set at the shed part of the insulator blank so that a fixed relative positional relationship is formed between the reference plate and the insulator blank; After the insulator blank is placed upright on the horizontal plane, it is scanned to obtain the top view point cloud data, thereby extracting the top view reference plate point cloud and the upper positioning surface point cloud; after the insulator blank is placed upside down on the horizontal plane, it is scanned to obtain the bottom view point cloud data, thereby extracting the bottom view reference plate point cloud and the lower positioning surface point cloud. Calculate the first centroid of the top-view reference plate point cloud and the second centroid of the bottom-view reference plate point cloud, and construct the initial alignment transformation matrix of the top-view point cloud and the bottom-view point cloud based on the first centroid and the second centroid; Based on the point clouds of the bottom and top reference plates, source point sets and target point sets are obtained. The objective function is to minimize the positional deviation between the source point set and the target point set after rigid rotation transformation and translation compensation. The stage transformation matrix is ​​obtained by iteratively solving the objective function. The target bottom reference plate point cloud is obtained by transforming the bottom reference plate point cloud using the stage transformation matrix. The thickness of the reference plate is used as the translation amplitude and the normal vector direction of the target bottom reference plate point cloud is used as the translation direction to construct the translation matrix. Construct the target transformation matrix based on the initial alignment transformation matrix, the stage transformation matrix, and the translation matrix; The target transformation matrix is ​​used to transform the lower positioning surface point cloud to obtain the target lower positioning surface point cloud; based on the upper positioning surface point cloud and the target lower positioning surface point cloud, the positioning surface structure height of the insulator blank is obtained.

2. The method for measuring the height of the positioning surface structure of a hybrid insulator blank according to claim 1, characterized in that, Calculate the first centroid of the top-view reference plate point cloud and the second centroid of the bottom-view reference plate point cloud, and construct the initial alignment transformation matrix of the top-view and bottom-view point clouds based on the first and second centroids, including: The three-dimensional coordinates of the first centroid are obtained based on the average of the three-dimensional coordinates of each point in the top-view reference plate point cloud; the three-dimensional coordinates of the second centroid are obtained based on the average of the three-dimensional coordinates of each point in the bottom-view reference plate point cloud. Based on the x-axis and y-axis coordinates of the first centroid and the x-axis and y-axis coordinates of the second centroid, calculate the rotation angle of the second centroid to the first centroid about the z-axis; Based on the z-axis coordinates of the first centroid and the second centroid, calculate the translation distance along the z-axis from the second centroid to the first centroid. The initial alignment transformation matrix for the top and bottom viewpoint clouds is constructed based on the rotation angle and translation distance.

3. The method for measuring the height of the positioning surface structure of the hybrid insulator blank according to claim 2, characterized in that, The three-dimensional coordinates of the first centroid Represented as: ， in, The first point cloud in the top-view reference board represents the first... The three-dimensional coordinates of each point; This represents the total number of points in the point cloud of the top-view reference board; The three-dimensional coordinates of the second centroid Represented as: ， in, The first point cloud in the bottom reference plate represents the... The three-dimensional coordinates of each point; This represents the total number of points in the point cloud of the bottom-view reference plate; The rotation angle from the second mass center to the first mass center about the z-axis Represented as: ， Translational distance along the z-axis from the second centroid to the first centroid Represented as: ， Initial alignment transformation matrix Represented as: 。 4. The method for measuring the height of the positioning surface structure of a hybrid insulator blank according to claim 1, characterized in that, Based on the point clouds of the bottom-view and top-view reference plates, source and target point sets are obtained. The objective is to minimize the positional deviation between the source point set and the target point set after rigid rotation transformation and translation compensation. An objective function is constructed, and the stage transformation matrix is ​​obtained by iteratively solving the objective function, including: Extract the first edge point set from the top-view reference plate point cloud and the second edge point set from the bottom-view reference plate point cloud; The first objective function is constructed by taking the second set of edge points as the source point cloud and the first set of edge points as the target point cloud. The objective is to minimize the sum of the distances between each target source point after coordinate transformation by the first rotation matrix and the first translation vector in the source point cloud and each target point in the target point cloud that is closest to each source point. Based on the values ​​of the translation along the x-axis, the translation along the y-axis, and the rotation angle around the z-axis before and after the coordinate transformation, the first boundary constraint condition is constructed. The L-BFGS-B algorithm is used to iteratively solve the first objective function based on the first boundary constraint condition until the difference between the values ​​of the first objective function after two iterations is less than a preset function threshold, or the projection gradient of the first rotation matrix and the first translation vector after the iteration is less than a preset gradient threshold, or the number of iterations reaches a preset number of iterations. The first rotation matrix and the first translation vector are then obtained, thus obtaining the first-stage transformation matrix.

5. The method for measuring the height of the positioning surface structure of the hybrid insulator blank according to claim 4, characterized in that, The steps for extracting the edge point set of the reference board point cloud include: Each point in the point cloud of the baseline board is used as the point to be detected; A radius search is used to obtain the neighborhood point set of the point to be detected, and the azimuth angle of each neighbor point in the neighborhood point set relative to the point to be detected is calculated. The azimuth angles of all the neighbor points are arranged from largest to smallest to obtain an azimuth angle sequence. The difference between each azimuth angle in the azimuth angle sequence and the next azimuth angle is calculated. If the maximum value of all the calculated differences is greater than or equal to a preset angle threshold, the point to be detected is determined to be an edge point. If the maximum value of all the calculated differences is less than the preset angle threshold, the point to be detected is determined to be an interior point. Specifically, when calculating the difference between each azimuth angle in the azimuth angle sequence and the next azimuth angle, for the last azimuth angle, the difference between that azimuth angle and the first azimuth angle in the azimuth angle sequence is calculated. The sum of.

6. The method for measuring the height of the positioning surface structure of a hybrid insulator blank according to claim 4, characterized in that, The first objective function is expressed as: ， in, Represents the first rotation matrix; Indicates the first translation vector; Indicates the second edge point set. One point; Indicates the first edge point set with The nearest point; Indicates the number of points in the second edge point cluster; Represents the Euclidean norm; The first boundary constraint condition is expressed as follows: ， in, This represents the amount of translation along the x-axis before and after the coordinate transformation; This indicates the lower bound of the translation along the x-axis; This represents the upper bound of the translation along the x-axis; This represents the amount of translation along the y-axis before and after the coordinate transformation; This indicates the lower bound of the translation along the y-axis. This represents the upper bound of the translation along the y-axis; Represents the rotation angle around the z-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the z-axis; This represents the upper bound of the rotation angle about the z-axis.

7. The method for measuring the height of the positioning surface structure of a hybrid insulator blank according to claim 4, characterized in that, After obtaining the first-stage transformation matrix, the following is also included: A first-stage transformed bottom reference plate point cloud is obtained by performing coordinate transformation on the bottom reference plate point cloud using a one-stage transformation matrix. The bottom view reference plate point cloud is used as the source point cloud and the top view reference plate point cloud is used as the target point cloud. The objective function is to minimize the sum of the vertical distances between each target source point after coordinate transformation by the second rotation matrix and the second translation vector and the tangent plane of the target point cloud that is closest to each source point. Based on the values ​​of the rotation angle around the x-axis, the rotation angle around the y-axis, and the translation along the z-axis before and after the coordinate transformation, a second boundary constraint condition is constructed. The L-BFGS-B algorithm is used to iteratively solve the second objective function based on the second boundary constraint condition until the difference between the values ​​of the second objective function after two iterations is less than a preset function threshold, or the projection gradient of the second rotation matrix and the second translation vector after the iteration is less than a preset gradient threshold, or the number of iterations reaches a preset number of iterations. The second rotation matrix and the second translation vector are obtained, thus obtaining the two-stage transformation matrix. The target bottom-view reference plate point cloud is obtained by transforming the bottom-view reference plate point cloud using the two-stage transformation matrix.

8. The method for measuring the height of the positioning surface structure of a hybrid insulator blank according to claim 7, characterized in that, The second objective function is expressed as: ， in, Indicates the second rotation matrix; Indicates the second translation vector; This represents the first transformation of the bottom-view reference plate point cloud. One point; Indicates the point cloud of the top-view reference board and The nearest point; express The unit normal vector at that location; Represents the dot product operation of vectors; The second boundary constraint condition is expressed as follows: ， in, Represents the rotation angle around the x-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the x-axis; This represents the upper bound of the rotation angle about the x-axis; Indicates the rotation angle around the y-axis before and after the coordinate transformation; This represents the lower bound of the rotation angle about the y-axis; This represents the upper bound of the rotation angle about the y-axis; This represents the translation along the z-axis before and after the coordinate transformation; This represents the lower bound of the translation along the z-axis. This represents the upper bound of the translation along the z-axis.

9. The method for measuring the height of the positioning surface structure of a hybrid insulator blank according to claim 8, characterized in that, Target transformation matrix Represented as: ， in, Represents the translation matrix; Represents the two-stage transformation matrix; Represents the first-stage transformation matrix; This represents the initial alignment transformation matrix.

10. The method for measuring the height of the positioning surface structure of a hybrid insulator blank according to claim 1, characterized in that, Based on the point cloud of the upper positioning surface and the point cloud of the target lower positioning surface, the structural height of the positioning surface of the insulator blank is obtained, including: The first average height of the upper positioning surface is obtained based on the average z-axis coordinate of each point in the point cloud of the upper positioning surface. The second average height of the lower positioning surface is obtained based on the average z-axis coordinate of each point in the point cloud of the target lower positioning surface. The positioning surface structure height of the insulator blank is obtained based on the difference between the first average height and the second average height.

Images