Spacecraft structure deformation geometric reconstruction technology based on PCL point cloud library

Abstract

The invention discloses a spacecraft structure deformation geometric reconstruction technology based on a PCL point cloud library, and the technology comprises the steps: 1, carrying out the calculation of the deformation of a spacecraft in a stress state through employing a finite element algorithm, so as to obtain a corresponding deformation result file; and 2, carrying out point cloud extraction on the deformation result file, and carrying out geometric reconstruction based on a PCL point cloud library to obtain a deformed geometric entity model. The invention provides a spacecraft structure deformation geometric reconstruction technology based on a PCL point cloud library. A deformation finite element grid is converted into watertight entity geometry through geometric reconstruction, the export format is the international STEP standard, structure modification and grid division can be conducted again through conventional engineering software such as machinery and mechanics, and a watertight three-dimensional geometric model is provided.

Description

Technical Field

[0001] The present invention relates to the field of geometric structure reconstruction, and more specifically, to a spacecraft structure deformation geometry reconstruction technology based on a PCL point cloud library. Background Art

[0002] During the flight of a spacecraft, it will deform under the influence of the external mechanical and thermal environment. The deformed structure will change the original aerodynamic effect, and then change the mechanical and thermal boundary, causing the flight trajectory to change. In order to more accurately predict the flight trajectory of the spacecraft structure, the structural model will deform after the structural mechanics calculation. The deformation result cannot be directly applied to the mechanical calculation at the next moment. The result needs to be converted into a new geometric model, that is, geometric reconstruction. The reconstructed model can be used for fluid mechanics and structural mechanics calculations in the next time period.

[0003] In the prior art, the following problems exist for geometric reconstruction of spacecraft:

[0004] 1) The existing technology mainly uses hardware such as cameras to scan the shape to obtain point clouds, and performs operations such as filtering, segmentation, sampling and registration on the point clouds. The final result is point cloud data that describes the shape of the object more accurately, but it is impossible to obtain accurate shape data for complex internal structures such as hollow shells. There is little further processing of the point cloud data such as curve and surface fitting and solid geometry generation.

[0005] 2) Existing technologies mainly operate at the point cloud extraction stage, combining different point cloud fragments through registration to obtain point cloud data with high signal-to-noise ratio. After that, most mature surface reconstruction work is mainly reconstructed in the form of 3D meshes. Although the fitted shape is visually close to the physical appearance, it is difficult to apply to further operations in engineering drawing software and mechanical analysis software. There are few reconstruction works that use point clouds as input objects and realize CSG and B-Rep;

[0006] 3) Existing technologies include 3D reconstruction of the environment based on the PCL point cloud library, which is used to reconstruct the surface contours of large-scale natural objects or architectural environments. Its accuracy is completely different from that of reconstruction in the mechanical field, because the reconstruction of the terrain does not require millimeter or higher accuracy. Therefore, this type of technology is not suitable for the geometric reconstruction of spacecraft, and it is difficult to achieve the geometric accuracy required for spacecraft mechanical analysis.

[0007] 4) Spacecraft geometry reconstruction belongs to reverse engineering, serving mechanical calculations, and requires extremely high geometric accuracy. Spacecraft usually have complex configurations and can be decomposed into many primitives, making it difficult to generate valid entities from point clouds at one time.

[0008] 5) The deformation of the spacecraft under the action of thermal force is small, which will form an irregular surface. It is impossible to obtain the deformed surface only by primitive fitting method, and it is also necessary to combine the spline fitting method; Summary of the invention

[0009] An object of the present invention is to solve at least the above problems and / or disadvantages and to provide at least the advantages which will be described hereinafter.

[0010] In order to achieve these purposes and other advantages according to the present invention, a spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library is provided, comprising:

[0011] Step 1: Use finite element algorithm to simulate the spacecraft under stress state to obtain the corresponding deformation result file;

[0012] Step 2: Extract the point cloud from the deformation result file, and then perform geometric reconstruction based on the PCL point cloud library to obtain the deformed geometric entity model.

[0013] Preferably, in step 1, the process of using a finite element algorithm to perform simulation calculation on a spacecraft under a force state is configured to include:

[0014] S10, establishing a general solid model of the spacecraft structure and then meshing it;

[0015] S11, the meshed structure is statically solved using finite elements to obtain the result file of the spacecraft structure after mesh deformation.

[0016] Preferably, in step 2, the point cloud extraction method is configured to include:

[0017] The point cloud data in the deformation result file is read through the PCL point cloud library, and in the reading process, the identifier in front of the coordinate point data segment is used as the starting point of the judgment, and the end symbol is used as the end point of the judgment;

[0018] The read point cloud coordinate information is saved as a PCD point cloud format file to complete the point cloud reading operation.

[0019] Preferably, in step 2, the geometric reconstruction process is configured to include:

[0020] S20, performing boundary recognition on the extracted point cloud;

[0021] S21, segmenting the point cloud after boundary recognition, and obtaining a plurality of corresponding regions with the boundary points as boundaries;

[0022] S22, performing curve fitting on the points on the boundary of the sub-regions to obtain a plurality of sidelines connected to each region, and connecting the end points of the plurality of sidelines to form a plurality of closed loops;

[0023] S23, performing surface fitting on each ring and its middle point, and obtaining a smooth surface larger than the original one by supplementing the residual fitting, so as to facilitate subsequent trimming;

[0024] S24, stitching the fitted surface;

[0025] S25, export STEP neutral file.

[0026] Preferably, in S20, the boundary solution of the point cloud is implemented by an alpha-shape algorithm, specifically a concavehull function in the PCL point cloud library.

[0027] Preferably, in S21, the point segmentation is a segmentation algorithm based on the basic data structure and local curvature provided in the PCL point cloud library, so as to segment the point cloud in the overall space into point clouds of each independent area;

[0028] Among them, in the point cloud curvature segmentation algorithm, the curvature threshold of the entire point cloud outer surface is solved, the curvature threshold is scanned, and the part greater than the curvature threshold λ is used as the segmentation boundary, and the maximum point in each interval greater than λ is used as the segmentation point;

[0029] The intersection points of different directions are used as the dividing points of each curve segment, and multiple corresponding regions are obtained with the dividing points as boundaries.

[0030] Preferably, in S22, the curve fitting is configured to use a cubic NURBS curve for fitting, and the endpoints of the fitted curve take a fourth-order repetition, and the internal nodes of each curve segment use a uniform parameter method;

[0031] In S23, the surface fitting is configured to adopt NURBS surface fitting.

[0032] Preferably, in S24, the surface stitching is to stitch each individually fitted surface into a closed shell using Open-Cascade technology to obtain a watertight geometric structure.

[0033] Preferably, in S25, according to the surface export interface included in the PCL library, the closed spline surface can be exported to the step geometry model grid through the Open-Cascade technology.

[0034] The present invention has at least the following beneficial effects:

[0035] 1) Accurately capture the deformation position of the spacecraft structure and obtain a visual deformation surface;

[0036] 2) It uses NURBS curve and surface fitting method instead of the mesh patch fitting method obtained by most point cloud reconstruction, which has the effect of B-Rep expression;

[0037] 3) Geometry reconstruction converts the deformed finite element mesh into a watertight solid geometry, and the export format is in the international STEP standard, which can be used for structural modification and meshing again through conventional engineering software;

[0038] 4) It can be applied to multi-field coupling calculation scenarios during the reentry of spacecraft. The calculation results of the structural finite element thermal response are used as the input of the geometric reconstruction algorithm, and the resulting geometric structure can be used as the input of the computational fluid dynamics meshing, thus forming a bridge for fluid-solid coupling calculation and achieving fluid-solid bidirectional coupling calculation.

[0039] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the spacecraft geometry reconstruction process of the present invention;

[0041] Figure 2 A finite element mesh diagram of a square spacecraft structure in one embodiment of the present invention;

[0042] Figure 3 For Figure 2 Point cloud image for node extraction;

[0043] Figure 4 For Figure 3 Point cloud after boundary extraction;

[0044] Figure 5 For Figure 4 Example of point cloud after segmentation;

[0045] Figure 6 For Figure 5 Example graph after edge fitting;

[0046] Figure 7 A schematic diagram of two curves forming a ring;

[0047] Figure 8 A schematic diagram of a ring formed by four curves;

[0048] Fig. 9 It is a schematic diagram of the surface after cube fitting;

[0049] Fig.10 For Fig. 9 Schematic diagram of the bevel effect;

[0050] Fig.11 The schematic diagram of the solid cross section after the cube is fitted;

[0051] Fig.12 This is a schematic diagram of the structure of a return capsule in another embodiment of the present invention;

[0052] Fig.13 for Fig.12 Schematic diagram of the cross-section structure of the mid-return capsule;

[0053] Fig.14 for Fig.12 Schematic diagram of the forces acting on the mid-return capsule;

[0054] Fig.15 Schematic diagram of the mesh after the return capsule is deformed under force;

[0055] Fig.16 For Fig.15 The point cloud image extracted from the deformation result;

[0056] Fig.17 For the general Fig.16 A solid cross-sectional view generated after geometric reconstruction using the present invention. DETAILED DESCRIPTION

[0057] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0058] The present invention is based on the PCL point cloud library research of computer graphics, and can reconstruct geometric models through deformation results. This process is reverse engineering, and can achieve geometric reconstruction of complex models such as large spacecraft. Therefore, the present invention proposes to perform geometric reconstruction based on the point cloud library, integrate this module into the laboratory's independent fluid mechanics calculation and thermal response calculation program, and finally obtain a calculation program that can achieve fluid-solid-thermal coupling in continuous time, establish a geometric reconstruction technology suitable for spacecraft structure deformation, and realize its application in the stress deformation of complex structures such as return capsules.

[0059] The Point Cloud Library is a large-scale cross-platform open source C++ programming library built on the basis of previous point cloud related research. It can implement a large number of general point cloud related algorithms and efficient data structures, involving point cloud acquisition, filtering, segmentation, registration, retrieval, feature extraction, recognition, tracking, surface reconstruction, visualization, etc.

[0060] Geometric reconstruction based on the point cloud library can realize fluid-solid thermal coupling calculation in continuous time, which will greatly improve the efficiency of fluid-solid thermal coupling calculation during the re-entry of large spacecraft into the atmosphere.

[0061] The deformation of the spacecraft caused by the force can be calculated by the structural finite element method, which contains the finite element mesh data information after the spacecraft is deformed, including the coordinate values ​​of each node, which is the form of point cloud. For the huge amount of data point cloud information, various operations can be performed through the PCL library.

[0062] The point cloud data in this study comes from the calculation results of structural mechanics. To extract the geometric parameters and shape of the reconstructed model, the model needs to be retrieved and the surface shape of the model needs to be fitted from the point cloud data. The kd-tree and oc-tree included in PCL can efficiently compress, store and manage massive point clouds. Filtering, feature description and advance basic processing can be applied to intelligent retrieval of models, surface reconstruction and visualization.

[0063] Specifically, if Figure 1 As shown, the spacecraft structure deformation geometry reconstruction technology based on the PCL point cloud library of the present invention is configured to include:

[0064] Step 1: using a finite element algorithm to calculate the deformation of the spacecraft under the stress state to obtain a corresponding deformation result file. In step 1, the process of analyzing the deformation of the spacecraft under the stress state using the finite element algorithm is configured to include:

[0065] S10, establishing a general solid model of the spacecraft structure and then meshing it;

[0066] S11, the meshed structural parts are statically solved using finite elements to obtain the deformation result file of the spacecraft structure after finite element analysis;

[0067] Step 2: Extract the point cloud from the deformation result file, and then reconstruct the geometry based on the PCL point cloud library to obtain the deformed solid model.

[0068] Furthermore, the main technical process of geometric reconstruction in step 2 of the present invention is configured to include:

[0069] S1, point cloud extraction

[0070] By using finite element correlation methods to solve the statics of structural parts, a deformation result file will be obtained, which contains the coordinate data of each grid node after deformation. These data are extracted to form a point cloud.

[0071] Usually, the deformation result file can be opened as a text file, which contains the file header field corresponding to the finite element calculation. There will be a data segment below the file header field, which contains the x, y, and z coordinate segments, which are the source of the point cloud data. The file interface is written in C++ language to read the point cloud data. During the reading process, the identifier in front of the coordinate point data segment is used as the criterion to start reading. The end mark after the coordinate point data segment indicates that the reading is completed, and the coordinate information of these points is saved as a PCD file, which is a file format dedicated to the PCL point cloud library. Take the tetrahedron finite element mesh file of a cube as an example, such as Figure 2 As shown, the cube grid model is used here for illustration without affecting the applicability of the structural deformation reconstruction technology. The point cloud displayed after reading is as follows Figure 3 As shown, the PCL point cloud library can efficiently manage point clouds.

[0072] S2, boundary point cloud recognition

[0073] The boundary of the point cloud is the original data of the fitting surface, and its extraction accuracy directly affects the final geometric reconstruction effect.

[0074] The boundary solution of the point cloud is based on the alpha-shape algorithm, which is based on the principle of rolling a circle with alpha as the radius around the boundary of the plane point cloud to gradually obtain boundary points. When extended to three-dimensional space, the point cloud boundary is still extracted based on the similar convex hull principle, specifically the concavehull function in the PCL point cloud library.

[0075] Among them, the alpha value is usually selected as 2.5 times the point cloud density, which can efficiently extract high-precision point cloud boundaries.

[0076] Among them, the point cloud density ρ refers to the distance d between each point in the point cloud and the nearest neighbor point k The average value of . Solving the nearest neighbor distance of each point will greatly increase the time cost of solving. The point cloud density required by this method does not require an extremely accurate value. When there are no abnormal outliers in the point cloud, its density will not deviate greatly due to incomplete sampling points. The sampling can be distributed, and the nearest neighbor distance d of the sampling point can be obtained through kd-tree k The density is obtained by the following formula to obtain the point cloud density ρ:

[0077]

[0078] In the above formula, m is the total number of points in the point cloud, n is the number of sampling points, and n s is the sampling interval, k = 1, 2, 3, … n.

[0079] The result after extracting the cube point cloud boundary is as follows Figure 4, the points in the original solid area are erased, and the outermost surface point cloud data is retained. If it is a closed shell with a certain thickness like the return capsule, then there will be an inner boundary and an outer boundary.

[0080] S3, point cloud segmentation

[0081] The PCL point cloud library provides basic data structures for point cloud segmentation and some general segmentation algorithms to segment the point cloud in the overall space into point clouds of each independent region so that the subsequent surfaces of each region can be independently fitted. The regional point cloud can form an independent surface, and the absence of cross surfaces meets the segmentation requirements. At this time, the segmentation point becomes the boundary point on the regional ring. The point cloud composition on each edge is fixed and is used by two adjacent surfaces at the same time. In the subsequent curve and surface fitting process, these points must be passed.

[0082] Since the segmented point cloud is on the surface or shell, a segmentation algorithm based on local curvature is used here. The point cloud here is obtained from the finite element analysis results. There is basically no noise and no noise reduction processing is required. The point cloud curvature segmentation algorithm is studied to solve the curvature λ of the entire point cloud outer surface, and the curvature threshold is scanned. Any part greater than the threshold λ is the segmentation boundary. In each interval greater than λ, the maximum point is used as the segmentation point. When the segmentation point sets in different directions intersect, they become the dividing points of each curve. When the segmentation points of the entire point cloud are found, the point cloud has been preliminarily separated. A bounding box is established with the ring where the segmentation point is located, and all points in the bounding box are points in this area.

[0083] like Figure 5 As shown in the figure, the outer surface of the cube is divided into 6 areas. At this time, the boundary line is composed of discrete points on a dividing edge. Curve fitting has not been performed yet. Multiple boundary lines a connected end to end can form a closed loop. The inside of each closed loop is a point on the same surface. The boundary line is a common line, and these points are also common points in the subsequent fitting process.

[0084] S5, boundary curve fitting of point cloud in different regions

[0085] By fitting the points on the boundary of the sub-regions, we can obtain the edge lines, which are the wireframe patterns in the graph. The endpoints of these edge lines are connected everywhere, and multiple edge lines form closed loops.

[0086] The cubic NURBS curve is used for fitting, and the fourth-order repetition is taken at the end points of the fitted curve to ensure that the curve passes through the two end points (points of the point cloud) and the adjacent curves form a seamless connection. The internal nodes of each curve segment use the uniform parameter method, because the point cloud of the fitting environment is the result of finite element calculation and has a certain uniformity. In order not to complicate the problem, the weight of the point value is taken as 1.

[0087] like Figure 6 As shown, after the cube is fitted, every 4 edge lines b can form a ring (the area where the arrow is located is a ring), and the 8 edges constitute a total of 6 rings, among which C is the fitted edge line.

[0088] S6, surface fitting of each ring

[0089] NURBS surface fitting is performed on each ring and its middle point respectively. Since the edges are common, the coordinates remain unchanged. After obtaining the surfaces of each ring, all of them can be combined together to form a closed surface. By supplementing the residual fitting, a smooth surface larger than the original one can be obtained for subsequent trimming.

[0090] In the fitting of a single surface, if multiple orders of repetition are not taken at the endpoints, a face group that exceeds the type value point will be obtained. In this case, it cannot be used normally, and the face group needs to be trimmed to make the point cloud just within the coverage of the fitted surface. If trimming is used, new problems will arise. Whether the trimmed surface A can maintain a seamless edge with the trimmed surface B with the same edge, this processing work will be greater. The method adopted by this patent uses the edge as a constraint. For example, the method of taking the 4th order repetition at the endpoint of the curve in the previous section strictly limits the fitting points of the surface, which can avoid the problem of broken surfaces and penetration of adjacent surfaces.

[0091] A closed loop consists of at least two curves. Usually, the number of type value points on any two lines is inconsistent, such as Figure 7 As shown in the figure, the curve Ⅰ1 has more type value points than curve Ⅱ2, so the curve Ⅱ2 with fewer points is supplemented with corresponding points, and then the first direction Ⅰ is fitted through the type value points in the middle of the surface. a For the curve fitting of the second direction Ⅰb, a curve Ⅷ3 can be fitted by selecting the middle point passing through the first direction. Together with curves Ⅰ1 and Ⅱ2, there are three curves in total. Combined with the characteristic mesh vertices c and d of the curve fitted in the first direction, they constitute the entire surface.

[0092] For Figure 8 In the case of the four curves shown (curve III4, curve V5, curve VI6, curve VII7, where the intersections of the curves have characteristic mesh vertices e, f, g, h), each direction is line-to-line, and the same direction directly takes the type value points with more points for fitting. When fitting in the second direction IIi, the number of type value points of the edge curve remains consistent, and the middle type value point must be the type value point passed by the first direction IIj. In the case of other different numbers of curves, if the point corresponds to the line, the point is reused and fitted to the corresponding line. If a line corresponds to multiple lines, this line is divided into multiple lines, and then the number of type value points is added or subtracted for fitting.

[0093] The surface fitting effect of the cube point cloud is as follows Fig. 9 , Fig.10 As shown, Fig. 9 For the combination of closed surfaces, Fig.10 For its beveled surface, it can be seen that there is only a face group.

[0094] S7, surface stitching and construction of STEP neutral file

[0095] When all the face groups are obtained, it is necessary to stitch together the relatively independent surfaces to form a watertight shell structure. Since the common edges of each surface have been obtained, there is no need to solve the intersection lines additionally. The BRepTools_Quilt class in the Open-Cascade technology can be used to stitch the connected surfaces into a shell. What is ultimately needed is the output step format file, which is the storage carrier of the B-Rep representation of the geometric model. The B-Rep representation is a boundary representation method, and the entity is surrounded by a finite number of faces. The geometric information of the faces, edge equations, and vertex coordinate values ​​that constitute the shape is recorded in detail, and the topological information between these geometric elements, that is, the composition relationship of the face vertices, is described. In the steps from point cloud to surface fitting, the information required for B-Rep expression is covered.

[0096] This part needs to detect whether the generated shell structure meets the watertightness condition. Only closed surfaces can generate solid structures. Two conditions are required:

[0097] 1. Each edge is shared by exactly two corresponding surfaces;

[0098] 2. Each vertex is shared by no less than 3 curves.

[0099] STEP is a product data exchange standard developed by a technical committee established by the International Organization for Standardization (ISO), which contains solid model information such as geometry, topology, and dimensions.

[0100] Finally, through the step interface integrated by the STEPControl_Writer class of Open-Cascade technology, the B-Rep entity can be converted into step format and output.

[0101] After the generated step model is imported into the 3D software, the model can be displayed and edited, and mechanical analysis can also be performed in other engineering software. Fig.11 To generate a cross-sectional rendering of a solid, the middle part is entirely a solid structure.

[0102] Examples of the reconstruction technology:

[0103] The invention is mainly used in spacecraft structural components, and the deformation of the return capsule is taken as an example. The return capsule is a typical curved surface structure with relatively complex surface parameters, and it is a hollow shell, which is extremely difficult to deform and reconstruct.

[0104] 1) The return capsule deforms under stress

[0105] Return capsule Fig.12 As shown, Fig.13 For its cross-section, the model has solid thickness.

[0106] Apply loads to the return capsule to cause it to deform, such as Fig.14 As shown, the bottom is fixed and point loads are applied on both sides.

[0107] 2) Extraction of deformed point cloud data

[0108] The deformation of the return capsule under stress is as follows: Fig.15 As shown in the figure, the structural finite element method is used for analysis, and the mesh is deformed. The mesh of the deformed result contains the data of all nodes. After extracting the point cloud separately, Fig.16 shown.

[0109] 3) Point cloud geometry reconstruction based on PCL

[0110] According to the operation method in the geometric reconstruction technology of this patent, Fig.16 By performing the operation, you can obtain a solid model in STEP format. Fig.17 From its cross-sectional view, it can be found that the regenerated entity after deformation still has thickness and can be processed again by CAD software and mechanical analysis software.

[0111] The above solution is only an illustration of a preferred embodiment, but is not limited thereto. When implementing the present invention, appropriate replacement and / or modification can be performed according to user needs.

[0112] The number of devices and processing scales described here are used to simplify the description of the present invention. Applications, modifications and variations of the present invention will be obvious to those skilled in the art.

[0113] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and the embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily realized. Therefore, without departing from the general concept defined by the claims and equivalent scope, the present invention is not limited to the specific details and the illustrations shown and described here.

Claims

1. A spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library, characterized in that: include: Step 1: Use finite element algorithm to simulate the spacecraft under stress state to obtain the corresponding deformation result file; Step 2: Extract the point cloud from the deformation result file, and then perform geometric reconstruction based on the PCL point cloud library to obtain the deformed geometric entity model.

2. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library according to claim 1, characterized in that: In step 1, the process of using the finite element algorithm to simulate the spacecraft under the force state is configured to include: S10, establishing a general solid model of the spacecraft structure and then meshing it; S11, the meshed structure is statically solved using finite elements to obtain the result file of the spacecraft structure after mesh deformation.

3. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library according to claim 1, characterized in that: In step 2, the point cloud extraction method is configured to include: The point cloud data in the deformation result file is read through the PCL point cloud library, and in the reading process, the identifier in front of the coordinate point data segment is used as the starting point of the judgment, and the end symbol is used as the end point of the judgment; The read point cloud coordinate information is saved as a PCD point cloud format file to complete the point cloud reading operation.

4. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library according to claim 1, characterized in that: In step 2, the geometric reconstruction process is configured to include: S20, performing boundary recognition on the extracted point cloud; S21, segmenting the point cloud after boundary recognition, and obtaining a plurality of corresponding regions with the boundary points as boundaries; S22, performing curve fitting on the points on the boundary of the sub-regions to obtain a plurality of sidelines connected to each region, and connecting the end points of the plurality of sidelines to form a plurality of closed loops; S23, performing surface fitting on each ring and its middle point, and obtaining a smooth surface larger than the original one by supplementing the residual fitting, so as to facilitate subsequent trimming; S24, stitching the fitted surface; S25, export STEP neutral file.

5. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library as claimed in claim 3, characterized in that: In S20, the boundary solution of the point cloud is implemented by an alpha-shape algorithm, specifically a concavehull function in the PCL point cloud library.

6. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library as claimed in claim 3, characterized in that: In S21, point segmentation is a segmentation algorithm based on the basic data structure and local curvature provided in the PCL point cloud library to segment the point cloud in the overall space into point clouds of each independent area; Among them, in the point cloud curvature segmentation algorithm, the curvature threshold of the entire point cloud outer surface is solved, the curvature threshold is scanned, and the part greater than the curvature threshold λ is used as the segmentation boundary, and the maximum point in each interval greater than λ is used as the segmentation point; The intersection points of different directions are used as the dividing points of each curve segment, and multiple corresponding regions are obtained with the dividing points as boundaries.

7. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library as claimed in claim 3, characterized in that: In S22, the curve fitting is configured to use a cubic NURBS curve for fitting, and the endpoints of the fitted curve take a fourth-order repetition, and the internal nodes of each curve segment use a uniform parameter method; in S23, the surface fitting is configured to use a NURBS surface fitting.

8. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library as claimed in claim 3, characterized in that: In S24, the surface stitching is to stitch each individually fitted surface into a closed shell using the Open-Cascade technology to obtain a watertight geometric structure.

9. The spacecraft structure deformation geometry reconstruction technology based on PCL point cloud library as claimed in claim 3, characterized in that: In S25, according to the surface export interface included in the PCL library, the closed spline surface can be exported to the step geometry model format through the Open-Cascade technology.

Images