Contour point structured processing method for airplane concave polygon thin-wall part

Abstract

The invention belongs to the technical field of point cloud data processing, and discloses a contour point structured processing method for an airplane concave polygon thin-wall part. All contour points are divided into arc points and straight line points based on local linear fitting residual errors, and the arc points and the straight line points are clustered; sorting the interiors of the arc clusters according to the sizes of polar angles, sorting the linear clusters according to coordinate projection, and completing local ordering in the clusters; on the basis of end point distance nearest search, the clusters are arranged in sequence, and ordering among the clusters is achieved; and performing spline curve fitting on the contour points subjected to global sorting, fitting discrete three-dimensional contour points into a smooth contour curve, and realizing structured processing on the contour points of the airplane concave polygon thin-wall part. According to the method, sorting modes based on angles and distances are combined through a mode of firstly classifying and then locally-globally sorting, so that structuring of contour points of the airplane concave polygon thin-wall part with a complex structure is realized, and the problem of local order disorder caused by a single sorting mode is avoided.

Description

Technical Field

[0001] This invention belongs to the field of contour point cloud data processing technology, and relates to a method for structuring contour points for concave polygonal thin-walled parts of aircraft. Background Technology

[0002] In aerospace, precision manufacturing, and other fields, many irregular thin-walled parts have complex concave polygonal contours, with shapes composed of geometric features such as straight lines and arcs. To obtain the contours of these parts and provide a reference for subsequent machining or assembly, 3D laser scanning is typically used to acquire a point cloud of the part, then edge points are extracted and fitted to obtain a complete boundary contour line. However, the contour points extracted from the point cloud are usually unstructured, meaning the arrangement of contour points is disordered. Before fitting, a sorting method is needed to form a structured sequence of contour points.

[0003] Xu Jinting et al., in their patent "A Method for Generating Parallel Tool Paths for CNC Machining Contours Based on Point Cloud Data," used a nearest-neighbor search method for contour point sorting. After specifying a starting point, they searched for the nearest point as the next sequential point, setting an angle threshold to limit the search direction. This iterative search continued until the starting point was found again, forming a structured closed-loop contour. Zhang Hui et al., in their patent "Workpiece Contour Extraction Method, Storage Medium, and Terminal Equipment," used an angle-based contour point structuring method. After decentering the contour point cloud, they calculated the angle value of the line connecting each point to the origin, and sorted the contour points based on the magnitude of the monotonic angle values. These two methods are commonly used contour point cloud sorting methods, offering advantages of simplicity and speed. However, when sorting complex concave polygon contours, they are prone to partial order disorder, making it difficult to achieve overall structuring of the contour points. Summary of the Invention

[0004] This invention addresses the problem of easily disordered partial ordering and difficulty in overall structuring during the sorting of contour points for complex contour parts of aircraft. It proposes a method for structuring contour points of concave polygonal thin-walled aircraft parts. This method achieves its function through a four-step process: First, it determines the type of points and performs clustering based on the residuals of linear fitting of neighborhood points; second, it sorts arc clusters and straight line clusters according to polar angle and projection axis respectively, completing local ordering within clusters; third, it achieves inter-cluster ordering based on endpoint nearest search; finally, it reconstructs the discrete contour point group into a structured smooth contour curve based on spline curve fitting, thus realizing the structuring of contour points for concave polygonal thin-walled aircraft parts.

[0005] The technical solution of the present invention:

[0006] A method for structuring contour points of concave polygonal thin-walled parts for aircraft includes the following steps:

[0007] Step 1: Classify contour points;

[0008] In the global coordinate system, there is a set of contour points P=[p1,p2,…,p...] for a concave polygonal thin-walled part of an aircraft. N For any point p in the contour points i =[x i , y i , z i ] Calculate the Euclidean distance between the point and other points in the global coordinate system XY plane, and select all points whose distance from the point is less than a distance threshold as neighborhood points, where the distance threshold is equal to the radius of the arc part in the contour;

[0009] Calculate the projection range of all neighboring points in the x and y directions of the global coordinate system. If the projection range in the x direction is smaller than the projection range in the y direction, swap the x and y coordinates of the neighboring points to avoid poor fitting results caused by the fitted line being perpendicular to the x-axis. Perform linear least squares fitting on the neighboring points to obtain the fitted line. Calculate the perpendicular distance from each neighboring point to the fitted line and use it as the fitting residual. Calculate the average value of the residuals and set a residual threshold. If the average value of the fitting residuals is greater than the residual threshold, the point is considered to be in a region with large curvature changes and is classified as an arc point; otherwise, it is classified as a straight line point. The residual threshold is equal to one-tenth of the radius of the arc portion in the contour.

[0010] DBSCAN is used to classify arc points and straight points in three-dimensional space. The clustering radius and minimum number of core points are set to obtain clusters of arc points and straight points respectively. The clusters of arc points and straight points are then merged to form a set containing multiple independent clusters. Each independent cluster is an arc or a straight line, and the attribute of each independent cluster is recorded as an arc or a straight line.

[0011] Step 2: Local sorting within the cluster;

[0012] The set containing multiple independent clusters obtained in step 1 is processed by sorting the contents of each independent cluster.

[0013] For the arc clusters obtained by clustering arc points, calculate the centroid coordinates of all points within the cluster in the XY plane. Subtract the centroid coordinates from the coordinates of each point in the cluster. Translate the arc cluster until the centroid coincides with the origin of the global coordinate system. Convert the coordinates of the translated point clusters to polar coordinates and sort them in ascending order according to the polar angle to obtain arc point clusters sorted counterclockwise.

[0014]

[0015] in, The polar angle corresponding to the i-th contour point within the cluster;

[0016] To avoid the sorting starting point being inside the arc, check the polar angle difference between all adjacent points to see if there are any sudden polar angle changes. Points with polar angle differences greater than a threshold are designated as breakpoints (idx). The sorting is then restarted from the breakpoint (idx+1), resulting in the following new order: ;

[0017] To achieve local ordering within a cluster of arc points, the polar coordinate system is restored to the Cartesian coordinate system and the translation is added to obtain the ordered arc point group in the original coordinate system.

[0018] For a cluster of straight lines, calculate the projection range of the cluster in the x and y directions in the global coordinate system. If the projection range in the x direction is greater than the projection range in the y direction, sort the points in the cluster in ascending order according to the x coordinate; otherwise, sort the points in the cluster in ascending order according to the y coordinate.

[0019] Step 3: Global inter-cluster sorting;

[0020] Let the set of clusters that have been locally ordered be . Define cluster C i The starting point is The destination is Initially, the first cluster index π(1) = 1 is selected, and a cluster set S is set for global ordering. The first cluster S(π(1)) = C1. At this time, there is only one cluster in the set S, and the last point p in the set is... cur The end point of the first cluster ;

[0021] Starting from the second cluster, iteratively find the distance p. cur The nearest intra-cluster start and end points, calculate p cur Euclidean distance to the two endpoints of all unsorted clusters, i.e., the start and end points. and Choose the cluster index corresponding to the minimum distance endpoint as ,like < If the order is not specified, the internal order of the cluster remains unchanged; otherwise, the point cloud of the cluster is arranged in reverse order. Add to the current sorted set S(k), update the current point as ;

[0022] Repeat the above process until all n clusters are sorted, resulting in an ordered set of clusters: This involves including all clusters, then releasing the points from each cluster and combining them according to the pre-arranged cluster order to obtain a structured, ordered set of points. ;

[0023] Step 4: Structured contour point spline curve fitting;

[0024] B-spline curve fitting point group To obtain a smooth and continuous profile curve, set the spline order m, the number of control points N+1, and construct the node vector U={u0,u1,⋯,u N+m The first m nodes are 0, the last m nodes are 1, and the internal nodes are uniformly distributed in the interval (0,1). The B-spline basis function N is calculated based on the Cox-de Boor recursive formula. i,m (u), calculate the coordinates of the curve at parameter u:

[0025]

[0026] In the node interval [u k-1 ,u N+1 Within a set step size, the fitted curve is sampled uniformly to obtain a continuous set of points on the fitted curve.

[0027] The beneficial effects of this invention are as follows: This invention effectively combines angle- and distance-based sorting methods by first classifying and then sorting locally and globally. This achieves the structuring of contour points of concave polygonal thin-walled parts of aircraft with complex structures, avoiding the problem of local order disorder caused by a single sorting method. It can be applied to the sorting of contour points of various parts in aircraft manufacturing. Attached Figure Description

[0028] Figure 1 This is a flowchart of the present invention.

[0029] Figure 2 This is a diagram illustrating the classification results.

[0030] Figure 3 This is a schematic diagram of the localized ordering of a cluster of circular arcs.

[0031] Figure 4 This is a schematic diagram showing the result after structuring the contour points of a certain part as a whole and fitting it with a spline curve. Detailed Implementation

[0032] The specific implementation of this method is further explained below with reference to the accompanying drawings and technical solutions.

[0033] Example 1

[0034] like Figure 1 As shown in the flowchart, a method for structuring contour points of concave polygonal thin-walled parts for aircraft includes the following steps:

[0035] (1) Classification of contour points

[0036] In the global coordinate system, there is a set of contour points P=[p1,p2,…,p...] for a concave polygonal thin-walled part of an aircraft. NFor any point p in the contour points i =[x i , y i , z i ] Calculate the Euclidean distance between the point and other points in the global coordinate system XY plane, and select all points whose distance from the point is less than a distance threshold as neighborhood points, where the distance threshold is equal to the radius of the arc part in the contour;

[0037] Calculate the projection range of all neighboring points in the x and y directions of the global coordinate system. If the projection range in the x direction is smaller than the projection range in the y direction, swap the x and y coordinates of the neighboring points to avoid poor fitting results caused by the fitted line being perpendicular to the x-axis. Perform linear least squares fitting on the neighboring points to obtain the fitted line. Calculate the perpendicular distance from each neighboring point to the fitted line, using it as the fitting residual. Calculate the average value of the residuals and set a residual threshold. If the average value of the fitting residuals is greater than the residual threshold, the point is considered to be in a region of large curvature change and is classified as an arc point; otherwise, it is classified as a straight line point. The residual threshold is equal to one-tenth of the radius of the arc portion in the contour. The results are as follows: Figure 2 As shown, the dots represent arc points, and the star points represent straight line points;

[0038] DBSCAN classification was performed on arc points and straight line points in 3D space. Clustering radii and minimum number of core points were set to obtain clusters for arc points and straight line points respectively. These clusters were then merged to form a set containing multiple independent clusters. Each independent cluster represents either an arc or a straight line, and the attribute of each cluster was recorded as either an arc or a straight line. The classification results are shown below. Figure 2 As shown.

[0039] (2) Local ordering within a cluster

[0040] The set containing multiple independent clusters obtained above is processed, and the internal order of each independent cluster is sorted.

[0041] like Figure 3 As shown, for the arc clusters obtained by clustering arc points, the centroid coordinates of all points within the cluster on the XY plane are calculated. The coordinates of each point in the cluster are subtracted from the centroid coordinates. The arc cluster is then translated until its centroid coincides with the origin of the global coordinate system. The coordinates of the translated point clusters are converted to polar coordinate system expressions and sorted in ascending order according to the polar angle to obtain arc point clusters sorted counterclockwise.

[0042]

[0043] in, The polar angle corresponding to the i-th contour point within the cluster;

[0044] To avoid the sorting starting point being inside the arc, check the polar angle difference between all adjacent points to see if there are any sudden polar angle changes. Points with polar angle differences greater than a threshold are designated as breakpoints (idx). The sorting is then restarted from the breakpoint (idx+1), resulting in the following new order: ;

[0045] This achieves local ordering within the cluster of arc points, then restores the polar coordinate system to the Cartesian coordinate system and adds translations to obtain the ordered arc point group in the original coordinate system;

[0046] For a cluster of straight lines, calculate the projection range of the cluster in the x and y directions in the global coordinate system. If the projection range in the x direction is greater than the projection range in the y direction, sort the points in the cluster in ascending order according to the x coordinate; otherwise, sort the points in the cluster in ascending order according to the y coordinate.

[0047] (3) Global inter-cluster sorting

[0048] Let the set of clusters that have been locally ordered be . Define cluster C i The starting point is The destination is Initially, the first cluster index π(1) = 1 is selected, and a cluster set S is set for global ordering. The first cluster S(π(1)) = C1. At this time, there is only one cluster in the set S, and the last point p in the set is... cur The end point of the first cluster ;

[0049] Starting from the second cluster, iteratively find the distance p. cur The nearest intra-cluster start and end points, calculate p cur Euclidean distance to the two endpoints of all unsorted clusters, i.e., the start and end points. and Choose the cluster index corresponding to the minimum distance endpoint as ,like < If the order is not specified, the internal order of the cluster remains unchanged; otherwise, the point cloud of the cluster is arranged in reverse order. Add to the current sorted set S(k), update the current point as ;

[0050] Repeat the above process until all n clusters are sorted, resulting in an ordered set of clusters: This involves including all clusters, then releasing the points from each cluster and combining them according to the pre-arranged cluster order to obtain a structured, ordered set of points. ;

[0051] (4) Structured contour point spline curve fitting

[0052] B-spline curve fitting point group To obtain a smooth and continuous profile curve, set the spline order m, the number of control points N+1, and construct the node vector U={u0,u1, ,u N+m The first m nodes are 0, the last m nodes are 1, and the internal nodes are uniformly distributed in the interval (0,1). The B-spline basis function N is calculated based on the Cox-de Boor recursive formula. i,m (u), calculate the coordinates of the curve at parameter u:

[0053]

[0054] In the node interval [u k-1 ,u N+1 Within a set step size, the fitted curve is uniformly sampled to obtain a continuous set of points on the fitted curve. The result of structuring the overall contour points of a part and fitting a spline curve is shown below. Figure 4 As shown.

[0055] The specific embodiments of the present invention have been described in detail above. The present invention proposes a structured processing method for contour points of concave polygonal thin-walled parts for aircraft. By linearly fitting the residuals of neighborhood points, the contour points are divided into arc and straight line portions. DBSCAN classification is used to form independent cluster sets. Polar angle sorting and projection sorting are used to sort the arc and straight line clusters within each cluster, respectively. Structured global contour points are searched by endpoint proximity. Spline curve fitting is applied to the discrete contour points to obtain a structured smooth contour, thus realizing the structured sorting of contour points for concave polygonal thin-walled parts of aircraft.

Claims

1. A method for structuring contour points of concave polygonal thin-walled parts for aircraft, characterized in that, Includes the following steps: Step 1: Classify contour points; In the global coordinate system, there is a set of contour points P=[p1,p2,…,p...] for a concave polygonal thin-walled part of an aircraft. N For any point p in the contour points i =[x i , y i , z i ] Calculate the Euclidean distance between the point and other points in the global coordinate system XY plane, and select all points whose distance from the point is less than a distance threshold as neighborhood points, where the distance threshold is equal to the radius of the arc part in the contour; Calculate the projection range of all neighboring points in the x and y directions of the global coordinate system. If the projection range in the x direction is smaller than the projection range in the y direction, swap the x and y coordinates of the neighboring points to avoid poor fitting results caused by the fitted line being perpendicular to the x-axis. Perform linear least squares fitting on the neighboring points to obtain the fitted line. Calculate the perpendicular distance from each neighboring point to the fitted line and use it as the fitting residual. Calculate the average value of the residuals and set a residual threshold. If the average value of the fitting residuals is greater than the residual threshold, the point is considered to be in a region with large curvature changes and is classified as an arc point; otherwise, it is classified as a straight line point. The residual threshold is equal to one-tenth of the radius of the arc portion in the contour. DBSCAN is used to classify arc points and straight points in three-dimensional space. The clustering radius and minimum number of core points are set to obtain clusters of arc points and straight points respectively. The clusters of arc points and straight points are then merged to form a set containing multiple independent clusters. Each independent cluster is an arc or a straight line, and the attribute of each independent cluster is recorded as an arc or a straight line. Step 2: Local sorting within the cluster; The set containing multiple independent clusters obtained in step 1 is processed by sorting the contents of each independent cluster. For the arc clusters obtained by clustering arc points, calculate the centroid coordinates of all points within the cluster in the XY plane. Subtract the centroid coordinates from the coordinates of each point in the cluster. Translate the arc cluster until the centroid coincides with the origin of the global coordinate system. Convert the coordinates of the translated point clusters to polar coordinates and sort them in ascending order according to the polar angle to obtain arc point clusters sorted counterclockwise. ； in, The polar angle corresponding to the i-th contour point within the cluster; To avoid the sorting starting point being inside the arc, check the polar angle difference between all adjacent points to see if there are any sudden polar angle changes. Points with polar angle differences greater than a threshold are designated as breakpoints (idx). The sorting is then restarted from the breakpoint (idx+1), resulting in the following new order: ; To achieve local ordering within a cluster of arc points, the polar coordinate system is restored to the Cartesian coordinate system and the translation is added to obtain the ordered arc point group in the original coordinate system. For a cluster of straight lines, calculate the projection range of the cluster in the x and y directions in the global coordinate system. If the projection range in the x direction is greater than the projection range in the y direction, sort the points in the cluster in ascending order according to the x coordinate; otherwise, sort the points in the cluster in ascending order according to the y coordinate. Step 3: Global inter-cluster sorting; Let the set of clusters that have been locally ordered be . Define cluster C i The starting point is The destination is Initially, the first cluster index π(1) = 1 is selected, and a cluster set S is set for global ordering. The first cluster S(π(1)) = C1. At this time, there is only one cluster in the set S, and the last point p in the set is... cur The end point of the first cluster ; Starting from the second cluster, iteratively find the distance p. cur The nearest intra-cluster start and end points, calculate p cur Euclidean distance to the two endpoints of all unsorted clusters, i.e., the start and end points. and Choose the cluster index corresponding to the minimum distance endpoint as ,like < If the order is not specified, the internal order of the cluster remains unchanged; otherwise, the point cloud of the cluster is arranged in reverse order. Add to the current sorted set S(k), update the current point as ; Repeat the above process until all n clusters are sorted, resulting in an ordered set of clusters: This involves including all clusters, then releasing the points from each cluster and combining them according to the pre-arranged cluster order to obtain a structured, ordered set of points. ; Step 4: Structured contour point spline curve fitting; B-spline curve fitting point group To obtain a smooth and continuous profile curve, set the spline order m, the number of control points N+1, and construct the node vector U={u0,u1, ,u N+m The first m nodes are 0, the last m nodes are 1, and the internal nodes are uniformly distributed in the interval (0,1). The B-spline basis function N is calculated based on the Cox-de Boor recursive formula. i,m (u), calculate the coordinates of the curve at parameter u: ； In the node interval [u k-1 ,u N+1 Within a set step size, the fitted curve is sampled uniformly to obtain a continuous set of points on the fitted curve.

Images