A method for measuring gaps and steps of an opening region of an aircraft skin based on a three-dimensional point cloud

Abstract

The application discloses a kind of based on three-dimensional point cloud's measurement method of gap and step difference of aircraft skin opening area, scanning hatch cover and skin measuring piece, obtains point cloud data and carries out pre-processing;Adjust point cloud space posture, the main direction of point cloud is aligned with global coordinate axis, make the space posture of point cloud data and global coordinate system fit, to provide uniform space reference for subsequent processing;Extract point cloud data profile;Transform source piece point cloud to target piece coordinate system, complete point cloud data virtual cooperation;Profile curve fitting;According to boundary point cloud data, its profile is fitted into smooth geometric curve;Calculate seam area gap and step difference.The application is based on the unified control of digital quantity reference, and the precision of seam area is predicted in advance by virtual assembly;Secondly, the application uses normal section projection method to calculate gap and step difference, which can be applied to any shape and large curvature cover test piece, and has good practicability.

Description

Technical Field

[0001] This invention belongs to the technical field of aircraft assembly and manufacturing, specifically relating to a method for measuring the gap and step difference of aircraft skin opening areas based on three-dimensional point clouds. Background Technology

[0002] In aircraft design and manufacturing, the aircraft's hatch is made of carbon fiber composite material, forming a mating relationship between the hatch and the skin with the docking structure. The gaps and steps in the joint area between the two have a crucial impact on the aircraft's aerodynamic and stealth performance. However, in actual production, the hatch and skin structures are manufactured independently before assembly. Although theoretical numerical models exist as a common dimensional coordination benchmark, dimensional coordination still needs to be performed according to the principle of mutual fitting during final assembly. Specifically, workers use feeler gauges or other tools to measure the gaps in the skin joint area and grind the edge allowances in place to ensure all gaps meet accuracy requirements. However, this method drastically increases the time cost of final assembly, and the in-place operation heavily relies on manual experience, easily leading to over-grinding damage to the panels. The reason for this is that the dimensional coordination in this process still relies on analog signal transmission, which obviously cannot meet the current stringent requirements for assembly efficiency and accuracy. Therefore, re-establishing a unified benchmark based on actual measurement data and coordinating dimensions based on digital transmission, so as to reduce the number of grinding operations or even eliminate the need for grinding during assembly, has become a core issue that urgently needs to be addressed in the assembly of skin caps.

[0003] Currently, the aerospace manufacturing industry has extremely strict requirements for the gaps and steps at the seams on the aircraft surface. Traditional measurement methods rely on manual labor and are inefficient, no longer meeting the current requirements for assembly efficiency and precision. Existing technologies mainly include the following three schemes for digital measurement of skin gaps and steps: (1) Parallel line structured light measurement method. This method has the advantages of high precision, low cost and good real-time performance, but it is more sensitive to the environment and has a limited measurement range. It is suitable for small-sized regular opening areas. For example, Chinese patent CN119832020A discloses a method for detecting gaps and steps in aircraft skin based on structured light. The method includes: acquiring skin seam images and original skin seam point clouds using binocular structured light; segmenting seam features on the skin seam images using a fine-tuned large-model deep learning network based on light compensation; mapping the seam features on the skin seam images to point cloud data; obtaining the skin seam pixel regions of the image and then performing error verification; then performing centerline fitting and profile line extraction to obtain the point cloud data distribution of the profile lines; finally, analyzing and calculating the gaps and steps of the obtained profile line point cloud data for edge chamfering. This method combines segmentation large-model analysis of two-dimensional images with multimodal point cloud processing using structured light, enabling rapid and accurate detection of gaps and steps in aircraft skin seams.

[0004] For example, Chinese patent CN115031646A discloses a method for expressing the step difference and gap model of aircraft skin seams based on line laser measurement. This method is used to assist in determining the type of aircraft skin seams when measuring the step difference and gap of aircraft skin seams using line laser measurement. The method is characterized by the following steps: 1) classifying aircraft skin seams into multiple structural types based on the seam size, seam shape, and whether there are chamfers or rounded corners; 2) constructing six seam expression models based on the different seam structures, using the endpoints of the light strip and their neighborhoods as expression elements, and using these models as input information for solving the step difference and gap of the skin seams.

[0005] (2) Image segmentation method. This method has the advantages of strong anti-interference ability and large range, but it has the disadvantages of relying on training dataset and being sensitive to fitting error. It is suitable for batch detection of medium-sized specimens. For example, existing Chinese patent CN119027339A discloses a method for removing noise from non-isolated point clouds based on image line laser stripe feature segmentation, including: 1 acquiring a line laser image of the object being scanned and extracting laser stripe features from it, then segmenting it to obtain a two-dimensional point set; 2 performing three-dimensional reconstruction on the two-dimensional point set to obtain three-dimensional point cloud data and performing feature analysis to obtain three-dimensional point cloud feature segmentation results; 3 determining and removing connected noise points in the non-isolated noise point cloud from the three-dimensional point cloud data based on the laser stripe features and the three-dimensional point cloud feature segmentation results, thereby converting the non-isolated noise point cloud into an isolated noise point cloud, and finally removing all isolated noise point clouds from the three-dimensional point cloud data.

[0006] (3) Point cloud scanning method has the advantages of large-scale measurement and strong robustness, but it is expensive for large-scale point cloud computing and depends on computing power. It is suitable for any size regular / irregular opening area.

[0007] For example, Chinese patent CN118865329A discloses a gap detection method based on 3D point cloud and deep learning, relating to the field of rail vehicle technology. The method includes: acquiring bogie image data for gap detection, and performing point identification on the bogie image data based on a target detection network to filter and obtain images to be processed; using 3D imaging technology to perform 3D reconstruction processing on the images to be processed to obtain point cloud data for gap measurement; performing semantic segmentation on the point cloud data for gap measurement based on a deep learning algorithm, and labeling the target region according to the boundaries between different semantic categories; using voxel-to-point cloud conversion technology to convert the target region into target point cloud data, and performing plane fitting processing on the target point cloud data to obtain geometric features for gap measurement, and calculating the bogie gap distance based on the geometric features.

[0008] For example, Chinese patent CN119515926A discloses a virtual assembly method for parts based on 3D point clouds. It includes the following steps: 1. Acquire point cloud data of the surface of the part under test; 2. Register the point cloud to the design model of the part; 3. Select a cross-section according to the position of the design model, extract the contour points of the point cloud on the cross-section, and calculate the manufacturing deviation of the part based on the contour information; 4. Plan the optimal assembly sequence with the minimum overall deviation based on the manufacturing deviation of each part; 5. Extract the cross-sectional contour perpendicular to the assembly surface from the point cloud of the part under the optimal assembly sequence, calculate the distance between the boundary points of the part, and output the assembly gap of the part.

[0009] Although the aforementioned digital measurement methods each have their own characteristics, the current coordination methods in the final assembly stage still rely on analog quantity transmission, which increases time costs and makes composite material panels prone to excessive wear. Summary of the Invention

[0010] The purpose of this invention is to provide a method for measuring the gap and step difference of aircraft skin opening areas based on three-dimensional point clouds. This method aims to address the problem in aircraft assembly and manufacturing where the final assembly stage relies on analog signal transmission for dimensional coordination, leading to increased time costs and the potential for over-grinding of composite material panels. This invention is particularly applicable to calculating the gap and step difference between aerospace composite material openings and mating skins.

[0011] This invention is mainly achieved through the following technical solutions: A method for measuring the gap and step difference in aircraft skin opening regions based on 3D point clouds includes the following steps: Step S1: Scan the cap and skin measuring parts to obtain point cloud data and perform preprocessing; Step S2: Adjust the spatial pose of the point cloud, align the main direction of the point cloud with the global coordinate axis, so that the spatial pose of the point cloud data is consistent with the global coordinate system, so as to provide a unified spatial reference for subsequent processing. Step S3: Extract the outline of the point cloud data; Step S4: Transform the source part point cloud into the target part coordinate system to complete the virtual matching of point cloud data; Step S5: Contour curve fitting; Based on the boundary point cloud data, fit its contour into a smooth geometric curve; Step S6: Calculate the gap and step difference in the joint area; Step S61: Use the normal vector section to cut out the joint area, and obtain two intersection points A( x 1, y 1, z 1) and B( x 2, y 2, z 2); Step S62: Place the two intersection points A ( x 1, y 1, z 1) and B( x 2, y 2, z 2) Along the normal vector The direction is projected onto the plane, and the Euclidean distance between the two projected points is calculated as the joint gap value. d ; Step S63: Place the two intersection points A ( x 1, y 1, z 1) and B( x 2, y 2, z 2) Along the normal vector Project the direction onto the tangent plane, and search for the nearest set of points in the neighborhood of the projection point. Calculate the step difference value for: ; ; ; in: h i The difference in height; w i As weight; The i-th nearest point found in the neighborhood of the projection point in the target point cloud; The i-th nearest point found in the projection point neighborhood of the source point cloud. For the i-th nearest point in the target point cloud The normal height coordinates; The i-th nearest point in the source point cloud The normal height coordinates.

[0012] To better implement the present invention, step S1 further includes the following steps: Step S11: Elevate the measuring parts of the cap and skin using the bracket, and scan to obtain their complete point cloud data; Step S12: Use a Gaussian filtering algorithm to remove noise points; Step S13: Downsample the point cloud data. Step S14: Use a surface interpolation algorithm based on radial basis functions to complete the locally missing data, and generate new point clouds by fitting implicit surfaces with neighborhood points.

[0013] To better realize the present invention, step S2 further includes the following steps: Step S21: First, move the centroid of the point cloud to the origin of the coordinate system; ; in: X This is the coordinate matrix of points in the original point cloud data; The coordinates of the centroid of the point cloud; X centered The coordinate matrix of the point cloud after the centroid is translated; Step S22: Calculate the covariance; ; in: Cov ( x Let be the covariance matrix of the point cloud coordinates; n is the total number of points in the point cloud; Step S23: Through Cov ( x Analysis points X , Y , Z The distribution characteristics of the direction are determined, and its degree of dispersion and the main distribution direction are obtained; Perform eigenvalue decomposition on the covariance matrix: ; in: v i These are the eigenvectors of the covariance matrix; l i For feature vectors v i The corresponding eigenvalues; eigenvalues l iThe direction of minimum value is taken as the main direction and serves as the reference direction for attitude adjustment; For example, if the three sets of eigenvalues ​​and their corresponding eigenvectors obtained are sorted in descending order of eigenvalues, let the sorted order of the eigenvalues ​​be denoted as . The corresponding feature vectors are as follows: v 1， v 2， v 3。 The smaller the eigenvalue, the more concentrated the point cloud distribution is in the corresponding direction. Therefore, the eigendirection corresponding to the smallest eigenvalue is selected as the reference main direction for attitude adjustment.

[0014] ; Step S24: Use the feature vector matrix V to rotate and adjust the point cloud pose so that the main direction of the point cloud is aligned with the global coordinate axis; ; in: X rotated It is the rotated point cloud coordinate matrix; V This is the eigenvector matrix.

[0015] To better realize the present invention, step S3 further includes the following steps: Step S31: First, search the neighborhood to find the target point. P i Centered on, within radius r Search for neighboring points within a range of no more than max_nn points to determine the local analysis range; ; in: P j Points within the neighborhood; max_nn is the threshold for the maximum number of points in the neighborhood; N ( i ) as the target point P i The set of neighborhood points; Step S32: Calculate the covariance to obtain the covariance matrix of the neighborhood points; perform eigenvalue decomposition on the covariance matrix, and take the eigenvector V3 corresponding to the smallest eigenvalue as the normal vector of that point. n i Then, make the normal vector n i All points towards the outer edge of the point cloud; Step S33: Does the proportion of neighborhood points with an angle greater than the normal vector exceed a certain threshold? sIdentify boundary points; ; ; in: i jk For points within the neighborhood P j and P k The angle between the normal vectors; i threshold The threshold value for the angle between the normal vectors; n j and n k Points P j and P k The normal vector; s This is the proportional threshold; For the set of neighborhood points N ( i Total points.

[0016] To better realize the present invention, step S4 further includes the following steps: Step S41: First, calculate the outer product of the normal vectors and the matrix H, and perform singular value decomposition on the outer product and matrix H to obtain the optimal rotation matrix R0; ; ; ; in: Let be the normal vector of the i-th point of the target component; Let i be the normal vector of the i-th point of the source component; U is a left singular vector matrix; V is a right singular vector matrix; It is a singular value diagonal matrix whose diagonal elements are matrix elements. H The singular values ​​are arranged in descending order; Step S42: Calculate the translation matrix t0 to align the centroid coordinates of the source and target parts through translation; Step S43: Construct a transformation matrix that includes rotation and translation to transform the source part point cloud into the target part coordinate system, thus completing the virtual assembly; ; Where T0 is the transformation matrix.

[0017] To better realize the present invention, step S5 further includes the following steps: Step S51: First, for the three line segment vertices A( on the contour) x 1, y 1) B ( x 2, y 2) and C( x 3, y 3) Use the cross product formula to determine which side of line AB point C is located on, thereby determining the concavity or convexity of the contour; ; like c If >0, then point C is outside line AB; if c If <0, then point C is inside line AB; if c If the line length is 0, then point C lies on line AB, meaning points A, B, and C are collinear. in: c The result of the cross product is used to determine the orientation of point C and line AB.

[0018] Step S52: Perform interpolation subdivision on the multi-line segment, generating n+1 subdivision points between adjacent vertices to divide the line segment into n segments, making the contour curve smoother.

[0019] The beneficial effects of this invention are as follows: (1) This invention obtains point clouds of the cap and skin through three-dimensional scanning. After noise reduction, completion and coordinate system alignment, the boundary contour is extracted based on the mutation of the normal vector. Virtual assembly is realized by normal vector matching. Finally, the gap is calculated by normal section projection method and the step difference is calculated by height difference weighting method. This invention significantly improves assembly efficiency and avoids damage to composite materials. It is applicable to various complex curvature structures.

[0020] (2) This invention uses the normal section projection method to calculate the gap and step difference, which is applicable to cover plate specimens of any shape and large curvature. Based on the unified control of digital reference, this invention predicts the accuracy of the joint area in advance through virtual assembly, replacing the traditional trial-and-error process of "in-situ grinding-repeated measurement", which greatly reduces the amount of manual grinding in the final assembly stage, and completely avoids the problem of excessive grinding damage to the wall plate due to insufficient manual experience. Through the establishment of the digital quantity transmission system, this invention realizes the transformation of the process paradigm from "experience-driven" to "data-driven", providing key technical support for the intelligent upgrading of the aerospace manufacturing field, and has good practicality. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the method for measuring the gap and step difference in aircraft skin opening regions based on three-dimensional point clouds, according to the present invention. Figure 2 A schematic diagram of the extracted contour for calculating the gap in the corresponding seam area; Figure 3 A schematic diagram of the extracted contour calculated for the corresponding step difference value; Figure 4 This is a schematic diagram of virtual assembly based on point cloud data; Figure 5 This is a schematic diagram after the contour curve has been fitted. Figure 6 A schematic diagram showing the projection of the intersection point of the joint area, which is obtained by using the normal vector section, onto the normal vector. Figure 7 This is a schematic diagram showing the calculation of the gap and step difference at any position after fitting in Embodiment 1 of the present invention. Detailed Implementation

[0022] Example 1: A method for measuring gaps and step differences in aircraft skin openings based on 3D point clouds, such as... Figure 1 As shown, it includes the following steps: Step 1: Determine the scanning platform and plan the scanning path; The scanning platform includes an intelligent optical tracking 3D scanner, a Track optical tracker, a robotic arm, a scanning platform, and a control module. The cover and skin measurement components are respectively elevated using supports to position them appropriately for scanning. The 3D scanner is mounted on the end effector of the robotic arm, and the robotic arm controller is operated to move along a predetermined path, achieving a comprehensive scan of the component and acquiring its complete point cloud data. If the aircraft skin is not removable, it must be scanned on-board. During the scanning process, the Track optical tracker tracks the laser emitter in real time. Because it is unaffected by relative position or vibration, it ensures that the measured point cloud does not jitter, guaranteeing the stability and accuracy of the point cloud data.

[0023] The aforementioned scanning system is compatible with both manual handheld and robotic arm automated modes. It only requires raising the measuring part with a support for scanning, without the need for complex tooling adjustments. It is suitable for carbon fiber composite caps and skin assemblies of different sizes and curvatures, and is significantly superior to the limitations of traditional measurement methods.

[0024] Step 2: Point cloud data preprocessing: point cloud data filtering, noise reduction, and simplification; Noise points are generated during the scanning process due to environmental interference and equipment errors. Therefore, a Gaussian filtering algorithm is used to remove noise points and improve the quality of the point cloud data. In addition, the point cloud data volume is enormous, and direct processing is computationally time-consuming. Therefore, downsampling is performed to simplify the point cloud data while preserving its feature data, thereby improving the efficiency of subsequent processing. Furthermore, for locally missing data, a surface interpolation algorithm based on radial basis functions (RBF) is used to complete the data, generating new point clouds by fitting implicit surfaces with neighborhood points.

[0025] Step 3: Point cloud spatial attitude adjustment; Based on the spatial distribution characteristics of point cloud data, the spatial attitude of the point cloud data is matched with the global coordinate system to facilitate subsequent virtual coordination.

[0026] First, move the centroid of the point cloud to the origin. ; in, X This represents the coordinate matrix of points in the original point cloud data. Represents the centroid coordinates of the point cloud. X centered This represents the coordinate matrix of the point cloud after the centroid has been translated.

[0027] Next, calculate the covariance: ; in, Cov ( x ) is the covariance matrix of the point cloud coordinates, and n is the total number of points in the point cloud.

[0028] By analyzing the distribution characteristics of points in the X, Y, and Z directions using this matrix, we can obtain their degree of dispersion and main distribution direction.

[0029] Then, eigenvalue decomposition is performed on the covariance matrix. ; in, v i These are the eigenvectors of the covariance matrix. l i It is an eigenvector v i The corresponding eigenvalues.

[0030] The main direction of the point cloud is determined based on the magnitude of the eigenvalues. The direction with the smallest eigenvalue is selected as the main direction because the point cloud distribution is most concentrated in this direction, and it can be used as the reference direction for attitude adjustment.

[0031] Finally, the eigenvector matrix V is used to rotate and adjust the point cloud pose, aligning the main direction of the point cloud with the global coordinate axis, thus providing a unified spatial reference for subsequent processing.

[0032] ; in, X rotated It is the rotated point cloud coordinate matrix. V This is the eigenvector matrix.

[0033] Step 4: Extraction of point cloud data contours; First, a neighborhood search is performed, starting with the current... P i Centered on a point, search for neighboring points within a radius of r and a neighborhood of no more than max_nn points to determine the local analysis range.

[0034] ; in, P i For the target point, P j For points within the neighborhood, The Euclidean distance between the two points is... r The search radius is denoted by `max_nn`, and `max_nn` is the threshold for the maximum number of points in the neighborhood. N ( i ) as the target point P i The set of neighborhood points.

[0035] Next, the covariance is calculated to obtain the degree of dispersion of the neighborhood points. ; ; in, m For neighborhood points N ( i The mean coordinates of ) For the set of neighborhood points N ( i The number of points in the matrix, where C is the covariance matrix of the neighborhood points. P j -μ represents the neighborhood point. P j Relative to the neighborhood center m The offset.

[0036] Then, eigenvalue decomposition is performed on the covariance matrix, and the eigenvector V3 corresponding to the smallest eigenvalue is taken as the normal vector of that point. n i .

[0037] ; ; Among them, V k λ represents the eigenvector of the neighborhood covariance matrix.k It is its corresponding eigenvalue.

[0038] To ensure that the normal vectors are consistent, if Then, the normal vector is inverted, so that all normal vectors point to the outside of the point cloud: n i =- n i ; in, n i For point P i The normal vector, μ- P i For the neighborhood center μ to P i The vector.

[0039] Next, calculate the angle between the normal vectors of each point in the neighborhood: ; in, i jk For points within the neighborhood P j and P k The angle between the normal vectors, n j and n k Points P j and P k The normal vector.

[0040] By judging the threshold of the included angle i threshold Does the proportion of the number of neighboring points exceed s To identify boundary points, when a sufficient number of points have normal vector changes exceeding a threshold, it can be determined that the point is located at the boundary of the point cloud. ; in, i threshold Let be the threshold for the angle between the normal vectors, and count be the number of points that satisfy the condition. s This is the proportional threshold. Let N(i) be the total number of points in the neighborhood point set N(i).

[0041] like Figure 2 As shown, two contours corresponding to the calculated gap values ​​of the joint area are extracted. Figure 3 As shown, two contours corresponding to the calculated seam area step difference are extracted.

[0042] Step 5: Virtual coordination of point cloud data; To simulate the assembly process of the lid and skin and achieve digital coordination, the outer product of the normal vectors and the matrix are calculated first.

[0043] ; Where H is the sum of the outer products of the normal vectors, Let be the normal vector of the i-th point of the target part. Let be the normal vector of the i-th point of the source component.

[0044] Then, singular value decomposition is performed on matrix H to obtain the optimal matrix.

[0045] ; Based on the decomposition results, determine the rotation matrix R0: ; Where R0 is the optimal rotation matrix.

[0046] Next, calculate the translation amount. ; The centroid coordinates of the source and target parts are as follows, and the centroids of the two parts are aligned by translation.

[0047] ; Finally, construct the transformation matrix that includes rotation and translation, such as... Figure 4 As shown, the point cloud of the source component is transformed into the coordinate system of the target component to complete the virtual assembly.

[0048] ; Where T0 is the transformation matrix.

[0049] Step 6: Contour curve fitting; Based on the boundary point cloud data, its contour is fitted into a smooth geometric curve. First, the vertices of the polyline segment are determined, taking three points A( on the contour) as an example. x 1, y 1) B ( x 2, y 2) and C( x 3, y 3) For example, use the cross product formula to calculate which side of line AB the determination point C is located on.

[0050] ; If the calculation result c If >0, then point C is outside line AB; if c If <0, then point C is inside line AB; if c If the line length is 0, then point C lies on line AB, meaning points A, B, and C are collinear, thus determining the concavity or convexity of the contour.

[0051] Then, interpolation subdivision is performed on the multi-segment, generating n+1 subdivision points between adjacent vertices using the following formula, dividing the segment into n segments. This makes the contour curve smoother and facilitates accurate calculation of gaps and step differences. The contour fitting result is as follows: Figure 5 As shown.

[0052] ; In the formula A ( x 1, y 1) B ( x 2, y 2) Using the coordinates of the two endpoints of the line segment to be subdivided; x i、 y i Let A be the coordinates of the starting point A of the line segment. x i= x 1， y i= y 1; n The number of segments into which a line segment is subdivided; j This is the index of the subdivision point, with a value from 0 to n, corresponding to the n+1 subdivision points generated; P j Let be the coordinates of the j-th subdivision point.

[0053] Step 7: Calculate the gap and step difference in the joint area.

[0054] like Figure 6 As shown, by using the normal vector section to cut the seam area, two intersection points A are obtained. x 1, y 1, z 1) and B( x 2, y 2, z 2), and along the normal vector Project the direction onto the plane, calculate the Euclidean distance between the two points after projection, which is the joint gap value d.

[0055] ; Along the normal vector Intersection point A ( x 1, y 1, z 1) and B( x 2, y 2, z 2) Project onto the tangent plane and search for the nearest point set in the neighborhood of the projected point. And calculate the height difference h i ; ; We calculate the order difference by weighting the values ​​based on the reciprocal of the distance. ; ; ; in, w i For weights.

[0056] like Figure 7 As shown, the final calculated joint gap value d and step difference value will be used to... They are displayed on the map respectively.

[0057] This invention employs a normal section projection method to calculate gaps and step differences, applicable to cover plate specimens of arbitrary shapes and large curvatures. Based on unified control using digital reference standards, this invention pre-determines joint area accuracy through virtual assembly, replacing the traditional trial-and-error process of "in-situ grinding-repeated measurement." This significantly reduces the amount of manual grinding work in the final assembly stage and completely avoids damage caused by excessive grinding of the panels due to insufficient human experience. Through the establishment of a digital quantity transmission system, this invention achieves a paradigm shift from "experience-driven" to "data-driven" processes, providing key technical support for the intelligent upgrading of the aerospace manufacturing field and demonstrating good practicality.

[0058] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications or equivalent changes made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for measuring the gap and step difference of an aircraft skin opening region based on three-dimensional point clouds, characterized in that, Includes the following steps: Step S1: Scan the cap and skin measuring parts to obtain point cloud data and perform preprocessing; Step S2: Adjust the spatial pose of the point cloud, align the main direction of the point cloud with the global coordinate axis, so that the spatial pose of the point cloud data is consistent with the global coordinate system, so as to provide a unified spatial reference for subsequent processing. Step S3: Extract the outline of the point cloud data; Step S4: Transform the source part point cloud into the target part coordinate system to complete the virtual matching of point cloud data; Step S5: Contour curve fitting; Based on the boundary point cloud data, fit its contour into a smooth geometric curve; Step S6: Calculate the gap and step difference in the joint area; Step S61: Use the normal vector section to cut out the joint area, and obtain two intersection points A( x 1, y 1, z 1) and B( x 2, y 2, z 2); Step S62: Place the two intersection points A ( x 1, y 1, z 1) and B( x 2, y 2, z 2) Along the normal vector The direction is projected onto the plane, and the Euclidean distance between the two projected points is calculated as the joint gap value. d ; Step S63: Place the two intersection points A ( x 1, y 1, z 1) and B( x 2, y 2, z 2) Along the normal vector Project the direction onto the tangent plane, and search for the nearest set of points in the neighborhood of the projection point. Calculate the step difference value for: ； ； ； in: h i The difference in height; w i As weight; The i-th nearest point found in the neighborhood of the projection point in the target point cloud; The i-th nearest point found in the projection point neighborhood of the source point cloud. For the i-th nearest point in the target point cloud The normal height coordinates; The i-th nearest point in the source point cloud The normal height coordinates.

2. The method for measuring the gap and step difference of aircraft skin opening regions based on three-dimensional point clouds according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Elevate the measuring parts of the cap and skin using the bracket, and scan to obtain their complete point cloud data; Step S12: Use a Gaussian filtering algorithm to remove noise points; Step S13: Downsample the point cloud data. Step S14: Use a surface interpolation algorithm based on radial basis functions to complete the locally missing data, and generate new point clouds by fitting implicit surfaces with neighborhood points.

3. The method for measuring the gap and step difference of aircraft skin openings based on three-dimensional point clouds according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: First, move the centroid of the point cloud to the origin of the coordinate system; ； in: X This is the coordinate matrix of points in the original point cloud data; The coordinates of the centroid of the point cloud; X centered The coordinate matrix of the point cloud after the centroid is translated; Step S22: Calculate the covariance; ； in: Cov ( x Let be the covariance matrix of the point cloud coordinates; n is the total number of points in the point cloud; Step S23: Through Cov ( x Analysis points X , Y , Z The distribution characteristics of the direction are determined, and its degree of dispersion and the main distribution direction are obtained; Perform eigenvalue decomposition on the covariance matrix: ； in: v i These are the eigenvectors of the covariance matrix; λ i For feature vectors v i The corresponding eigenvalues; Step S24: Using the eigenvector matrix V Complete the rotation adjustment of the point cloud attitude so that the main direction of the point cloud is aligned with the global coordinate axis; ； in: X rotated It is the rotated point cloud coordinate matrix; V This is the eigenvector matrix.

4. A method for measuring the gap and step difference of an aircraft skin opening region based on a three-dimensional point cloud, as described in claim 1 or 3, characterized in that... Step S3 includes the following steps: Step S31: First, search the neighborhood to find the target point. P i Centered on, within radius r Search for neighboring points within a range of no more than max_nn points to determine the local analysis range; ； in: P j Points within the neighborhood; max_nn is the threshold for the maximum number of points in the neighborhood; N ( i ) as the target point P i The set of neighborhood points; Step S32: Calculate the covariance to obtain the covariance matrix of the neighborhood points; perform eigenvalue decomposition on the covariance matrix, and take the eigenvector V3 corresponding to the smallest eigenvalue as the normal vector of that point. n i Then, make the normal vector n i All points towards the outer edge of the point cloud; Step S33: Does the proportion of neighborhood points with an angle greater than the normal vector exceed a certain threshold? σ Identify boundary points; ； ； in: θ jk For points within the neighborhood P j and P k The angle between the normal vectors; θ threshold The threshold value for the angle between the normal vectors; n j and n k Points P j and P k The normal vector; σ This is the proportional threshold; For the set of neighborhood points N ( i Total points.

5. The method for measuring the gap and step difference of an aircraft skin opening region based on three-dimensional point clouds according to claim 4, characterized in that, Step S4 includes the following steps: Step S41: First, calculate the outer product of the normal vectors and the matrix H, and perform singular value decomposition on the outer product and matrix H to obtain the optimal rotation matrix R0; ； ； ； in: Let be the normal vector of the i-th point of the target component; Let i be the normal vector of the i-th point of the source component; U is a left singular vector matrix; V is a right singular vector matrix; It is a singular value diagonal matrix whose diagonal elements are matrix elements. H The singular values ​​are arranged in descending order; Step S42: Calculate the translation matrix t0 to align the centroid coordinates of the source and target parts through translation; Step S43: Construct a transformation matrix that includes rotation and translation to transform the source part point cloud into the target part coordinate system, thus completing the virtual assembly; ； Where T0 is the transformation matrix.

6. The method for measuring the gap and step difference of an aircraft skin opening region based on a three-dimensional point cloud, as described in claim 5, is characterized in that... Step S5 includes the following steps: Step S51: First, for the three line segment vertices A( on the contour) x 1, y 1) B ( x 2, y 2) and C( x 3, y 3) Use the cross product formula to determine which side of line AB point C is located on, thereby determining the concavity or convexity of the contour; ； like γ If >0, then point C is outside line AB; if γ If <0, then point C is inside line AB; if γ If the line length is 0, then point C lies on line AB, meaning points A, B, and C are collinear. in: γ The result of the cross product is used to determine the orientation of point C and line AB; Step S52: Perform interpolation subdivision on the multi-line segment, generating n+1 subdivision points between adjacent vertices to divide the line segment into n segments, making the contour curve smoother.

Images