Implicit generation method for thin-wall lattice sandwich structure
Through the implicit generation method of thin-wall lattice sandwich structure, the problem of time-consuming and complex explicit generation and insufficient implicit generation accuracy is solved, and efficient and accurate thin-wall lattice sandwich structure modeling is achieved, which is suitable for cross-scale model filling.
Patent Information
- Application Number
- CN202510565695.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-30
- Publication Date
- 2025-08-26
AI Technical Summary
When generating thin-walled dot matrix sandwich structures, explicit generation technology is time-consuming and complex and has too sharp features, while implicit generation technology is difficult to deal with cross-scale and high-precision generation scenarios, resulting in low modeling efficiency and insufficient precision.
The implicit generation method of thin-wall dot matrix sandwich structure is adopted to generate a high-quality thin-wall model sandwich structure by dividing hexahedral mesh, voxelization processing, structuring distance field along the shape, and implicit Boolean fusion. The efficiency and accuracy are improved by combining GPU parallel computing.
It realizes efficient generation of thin-walled lattice sandwich structures, maintains model integrity and accuracy, avoids reconstruction errors of traditional methods, is suitable for cross-scale model filling, and can smoothly transition rounded corners.
Smart Images
Figure CN120541981A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of lightweight design, and in particular relates to an implicit generation method of a thin-walled lattice sandwich structure. Background Art
[0002] Lattice structures are periodically arranged porous structures that offer exceptional properties such as lightweighting, high specific stiffness, high energy absorption, and impact resistance. Lattice structure design is currently a key method for lightweighting in additive manufacturing and is widely used in aerospace, automotive, marine, and biomedical applications.
[0003] Currently, lattice microstructure filling has a wide range of applications, especially in thin-walled lattice sandwich structures, which require large lattice sizes and distinct cross-scale features. The mainstream explicit lattice generation technique, Boolean fusion, is complex and time-consuming, resulting in overly sharp features. Implicit generation techniques struggle to cope with high-precision generation across scales. Therefore, a lattice design method that achieves efficient modeling while balancing accuracy and integrity is urgently needed. Summary of the Invention
[0004] To address the aforementioned shortcomings and improvements in existing technologies, a method for implicitly generating thin-walled lattice sandwich structures is proposed. This method can stably generate hexahedral lattice-based filling structures within thin-walled structures. This thin-walled interior filling method, centered around lattice microstructures, efficiently achieves high-quality filling of interlayer microstructures across large, thin-walled models. This method addresses the limited precision of current mainstream implicit filling methods and the difficulty of merging and sharp transitions associated with explicit filling methods.
[0005] To solve the above problems, the present invention provides a method for implicitly generating a thin-walled lattice sandwich structure, and adopts the following technical solutions:
[0006] A method for implicitly generating a thin-walled lattice sandwich structure comprises the following steps:
[0007] Step 1: Divide the area to be filled in the thin-walled model and generate a hexahedral mesh that is geometrically conformal to the area to be filled, with the hexahedral mesh unit size consistent with the target lattice unit cell size;
[0008] Step 2: Voxelize the thin-walled model, select the boundary voxels that intersect with the model surface based on the separating axis theorem, and store the corresponding triangular facet information of the model in the boundary voxels;
[0009] Step 3: In each hexahedral grid cell, a conformal distance field is constructed by isoparametric transformation. The conformal distance field is generated by combining the lattice type and design parameters of the current hexahedral grid and the lattice information of the adjacent hexahedral grids.
[0010] Step 4: Based on the axis-aligned bounding box of each hexahedral mesh, the search range is expanded to cover the adjacent boundary voxels, the triangular facets stored in the adjacent boundary voxels are extracted, and the closest distance value from the sampling point to the associated triangular facet within the conformal region of each hexahedral mesh unit is calculated to form the local distance field of the thin-walled model, and the inner wall distance field of the model is generated through implicit bias;
[0011] Step 5: Perform implicit Boolean fusion on the lattice distance field and the inner wall distance field to generate a fused conformal distance field;
[0012] Step 6: Using the marching cube algorithm (MCM) to extract the isosurface from the conformal distance field, and outputting the STL model manufacturing file with the lattice and the inner wall fused together;
[0013] Step 7: Merge the triangular facet information of the original thin-wall model, the lattice and the inner wall fused STL model manufacturing file to form a complete thin-wall lattice sandwich structure model manufacturing file.
[0014] Furthermore, the hexahedral mesh division method in step 1 is: using the central axis algorithm to generate a hexahedral mesh that is geometrically conformal to the thin-walled model; and the division thickness h2 of the hexahedral mesh satisfies the following relationship:
[0015] h1 <h2<h0
[0016] Where h0 is the total thickness of the thin wall, and h1 is the actual thickness of the lattice filling area.
[0017] Furthermore, the specific steps of voxelization processing in step 2 include:
[0018] S1, axis-aligned bounding box based on thin-wall model, dividing uniform voxel grid in 3D space;
[0019] S2. For each triangular facet, filter the candidate voxels covered by its bounding box and use the separating axis theorem to detect the intersection between the voxel and the facet;
[0020] S3. Mark the intersecting voxels as boundary voxels and store the corresponding triangle patch indexes. After all patches are processed, the complete boundary voxel information of the thin-wall model is obtained.
[0021] Furthermore, the method for constructing the conformal lattice distance field in step 3 includes:
[0022] S1. Mapping the mesh vertices of the hexahedral mesh unit from the local coordinate system of the lattice unit cell to the global coordinate system of the model to be filled by isoparametric transformation to construct a conformal distance field;
[0023] S2. When generating a conformal lattice distance field based on the lattice type and design parameters of the current hexahedral mesh cell and the lattice information of the adjacent hexahedral mesh cells, the lattice parameters of up to 26 adjacent hexahedral mesh cells are incorporated to ensure geometric continuity at the boundary.
[0024] Furthermore, the search range in step 4 is expanded as follows:
[0025] S1. Based on the axis-aligned bounding box of the hexahedron, the expansion range along each coordinate axis is H = Δh + r node , get the search bounding box, where Δh = (h0-h1) / 2 is the inner wall offset thickness, r node is the maximum corner radius of the lattice node;
[0026] S2. Extract boundary voxels and their corresponding triangles from the voxels covered by the expanded search bounding box.
[0027] Furthermore, the formula for implicit Boolean fusion in step 5 is:
[0028]
[0029] Among them, f lattice (x) is the lattice distance field, which includes the lattice parameters of the current hexahedral grid unit and its adjacent units; L is the number of lattice cells, including up to 26 adjacent cells of the current lattice cell; d(li,x) is the closest distance from the sampling point x in the current hexahedral grid conformal region to the surface of the lattice cell li; f bound (x) is the inner wall distance field, T is the set of boundary triangles obtained by searching the bounding box area, d(ti,x) is the closest distance from the sampling point x in the current hexahedral mesh conformal area to the surface of the boundary triangle ti; s(x) is the sign function, if x is -1 inside the model and 1 outside the model, it is determined by the pseudo-normal method; Δh is the inner wall offset thickness; the fusion process introduces the fusion fillet radius r blend To eliminate sharp edges.
[0030] Furthermore, steps 3 to 6 are executed in parallel in batches on the GPU, specifically including:
[0031] S1. Divide the hexahedral mesh into preset batches based on the GPU hardware computing power, where each batch contains a preset number of hexahedral mesh units.
[0032] S2. Allocate multiple GPU thread blocks to each hexahedral grid cell in each batch, and use threads in the multiple GPU thread blocks to parallely calculate distance field values for all sampling points in the corresponding hexahedral grid cell.
[0033] S3. Execute steps 3 to 6 independently for each batch of hexahedral mesh elements and output them to the same manufacturing file;
[0034] S4. After all batches are processed in a loop, they are finally merged to generate an STL model manufacturing file with a complete fusion of the dot matrix and the inner wall.
[0035] Compared with the prior art, the present invention has the following beneficial effects:
[0036] The implicit generation method of thin-walled lattice sandwich structures provided by the present invention can be applied to any closed cavity lattice microstructure filling scenario including thin-walled sandwich structures; the required lattice generation accuracy can be controlled by controlling the resolution of the hexahedron conformal distance field, and it is applicable to any lattice type that can be implicitly expressed based on the hexahedron; compared with explicit modeling, it can efficiently achieve smooth fusion of the lattice and the inner wall of the skin, and the transition radius can be parameterized and controlled; compared with traditional implicit modeling methods, it can handle cross-scale model lattice filling tasks without losing accuracy; and the implicit generation method of thin-walled lattice sandwich structures described in the present invention only outputs the closed structure connected by the inner wall and the lattice, and forms a complete model after combining with the thin-wall original model structure, without changing the detailed features of the original model, thereby avoiding the reconstruction error of the traditional implicit method. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Figure 1 This is a flow chart of the method for generating a thin-walled lattice sandwich structure according to the present invention;
[0038] Figure 2 A two-dimensional schematic diagram of dividing the thin-walled area to be filled according to the present invention into hexahedral grid units;
[0039] Figure 3 A schematic diagram of the process of searching for boundaries in the model boundary voxels by the hexahedron conformal unit cell of the present invention;
[0040] Figure 4 Output the inner wall and lattice fusion model and the complete thin-wall lattice sandwich model example diagram for the present invention;
[0041] Figure 5 Schematic diagram of hexahedral division of the thin-walled area to be filled in an embodiment of the present invention;
[0042] Figure 6 Schematic diagram of a portion of the thin-walled lattice sandwich structure generation process according to an embodiment of the present invention. DETAILED DESCRIPTION
[0043] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0044] like Figures 1 to 4 As shown, the implicit generation method of the thin-walled lattice sandwich structure provided by the present invention includes the following steps:
[0045] Step 1: Divide the area to be filled in the thin-walled model and use the central axis algorithm to generate a hexahedral mesh that is geometrically conformal to the area to be filled. The unit size of the hexahedral mesh is consistent with the unit cell size of the target lattice, and each hexahedral mesh unit has a maximum of 26 adjacent units. To ensure that there is a certain fusion space between the lattice in the boundary hexahedron and the boundary area, the division thickness h2 of the hexahedral mesh satisfies the following relationship:
[0046] h1 <h2<h0
[0047] Among them, h0 is the total thickness of the thin wall, and h1 is the actual thickness of the lattice filling area;
[0048] Step 2: Based on the axis-aligned bounding box of the thin-walled model, divide the three-dimensional space into a uniform voxel grid. For each triangular facet, screen the candidate voxels covered by its bounding box and use the separating axis theorem to detect the intersection of the voxel and the facet. Mark the intersecting voxels as boundary voxels and store the corresponding triangular facet index. Once all facets are processed, the complete model boundary voxel information is obtained.
[0049] Step 3: Map the mesh vertices of the hexahedral mesh unit from the local coordinate system of the lattice unit cell to the global coordinate system of the model to be filled by isoparametric transformation to construct a conformal distance field; combine the lattice type and design parameters of the current hexahedron and the lattice information of adjacent hexahedrons, and incorporate the lattice parameters of up to 26 adjacent hexahedrons to generate a conformal lattice distance field;
[0050] This step is completed in parallel on the GPU;
[0051] Step 4: Based on the axis-aligned bounding box of each hexahedron, the range along each coordinate axis is expanded to H = Δh + r node , get the search bounding box, where Δh = (h0-h1) / 2 is the inner wall offset thickness, that is, the thickness from the inner wall to the outer wall of the filled model, r node is the maximum corner radius of the lattice node; extract the boundary voxels and the corresponding triangle facet information from the voxels covered by the search bounding box; use the obtained boundary information to calculate the closest distance value from the sampling point to the associated triangle facet within each hexahedral mesh conformal region to form the local distance field of the thin-walled model, and generate the model inner wall distance field by offsetting the thickness of Δh;
[0052] This step is completed in parallel on the GPU;
[0053] Step 5: Perform implicit Boolean fusion on the lattice distance field and the inner wall distance field to generate a fused conformal distance field. The specific formula is:
[0054]
[0055] Among them, f lattice (x) is the lattice distance field, which includes the lattice parameters of the current hexahedral grid unit and its adjacent units; L is the number of lattice cells, including up to 26 adjacent cells of the current lattice cell; d(li,x) is the closest distance from the sampling point x in the current hexahedral grid conformal region to the surface of the lattice cell li; f bound (x) is the inner wall distance field, T is the set of boundary triangles obtained by searching the bounding box area, d(ti,x) is the closest distance from the sampling point x to the surface of the boundary triangle ti in the current hexahedral mesh conformal area; s(x) is the sign function, if x is -1 inside the model and 1 outside the model, it is determined by the pseudo-normal method; Δh is the inner wall offset thickness; the fusion process introduces the fusion fillet radius r blend To eliminate sharp edges;
[0056] This step is completed in parallel on the GPU;
[0057] Step 6: Use the marching cube algorithm (MCM) to extract the isosurface from the conformal distance field and output the STL model manufacturing file with the lattice and the inner wall fused together. This step is completed in parallel on the GPU.
[0058] Step 7: Add the triangle facet information of the original thin-wall model to the STL model manufacturing file output in step 6 to form a complete thin-wall lattice sandwich structure model manufacturing file, such as Figure 4 shown.
[0059] Furthermore, according to the hardware computing power, steps 3 to 6 are executed in batches for the hexahedral grid units, and the inner wall lattice patches generated for each batch are output to the same model manufacturing file; after all batches are completed, they are merged to generate an STL model manufacturing file with a complete fusion of the lattice and the inner wall.
[0060] Example
[0061] The technical solution of the present invention is further explained below by taking a thin-walled wing sandwich filling structure as an example.
[0062] Thin-walled wing model:
[0063] Bounding box size: 407mm×284mm×110mm;
[0064] Total thickness of thin wall (h0): 2mm;
[0065] Filling area thickness (h1): 1.8mm;
[0066] Lattice type: double-layer BCC (body-centered cubic) structure;
[0067] Dot rod diameter: 0.15mm;
[0068] Inner wall offset thickness (Δh): 0.2mm;
[0069] Blend fillet radius (r blend ): 0.05mm;
[0070] Hardware and computing configuration:
[0071] Computational resolution: The unit cell is a conformal region of 32×32×32 sampling points;
[0072] Parallel strategy: 64 GPU thread blocks process a hexahedral unit cell, each thread block is assigned 512 computing threads, and each thread calculates the distance field value of a sampling point.
[0073] like Figure 5 and 6 As shown, the implicit generation method of the lattice sandwich structure of the thin-walled wing model includes the following steps:
[0074] S1-S2, using the central axis algorithm to generate a hexahedral mesh conformal to the wing surface, filled with a double-layer BCC lattice structure. The hexahedral mesh unit size is consistent with the BCC unit cell period (1.8mm×1.8mm×1.8mm), and the hexahedral partition thickness is slightly larger than the filling area thickness. At the same time, the thin-walled model is divided into voxel space and the boundary voxel information is extracted.
[0075] S3-S5, process hexahedral mesh cells in batches. For each batch of hexahedral mesh cells, map the hexahedral mesh cells to the global coordinate system through isoparametric transformation to ensure the geometric consistency of the BCC lattice in the curved surface area; when calculating the BCC distance field of the current unit cell, include the parameters of the surrounding 26 adjacent units to eliminate boundary mutations; search for boundary patches in the voxel space of the thin-walled model through the hexahedral offset bounding box, with the offset distance slightly larger than the distance from the inner wall to the outer wall (0.2mm); calculate the distance field from the sampling point to the associated patch in the conformal region of the hexahedral mesh, offset inward by 0.2mm to generate the inner wall distance field; perform implicit Boolean fusion on the lattice distance field and the inner wall distance field, with the fusion fillet radius set to 0.05mm;
[0076] S6-S7. After the fusion is completed, enter the parallel MCM to extract the isosurface triangles and output them to a unified STL file; after all batches are completed, output the complete inner wall and lattice fusion model file; add the triangular face information of the original thin-wall model to the output inner wall and lattice fusion model file to realize the construction of a complete thin-wall lattice sandwich structure.
[0077] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made by any person skilled in the art within the technical scope disclosed in the present invention and based on the technical solution and inventive concept of the present invention shall be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.
Claims
1. A method for implicitly generating a thin-walled lattice sandwich structure, characterized in that: The following steps are involved: Step 1: Divide the area to be filled in the thin-walled model and generate a hexahedral mesh that is geometrically conformal to the area to be filled, with the hexahedral mesh unit size consistent with the target lattice unit cell size; Step 2: Voxelize the thin-walled model, select the boundary voxels that intersect with the model surface based on the separating axis theorem, and store the corresponding triangular facet information of the model in the boundary voxels; Step 3: In each hexahedral grid cell, a conformal distance field is constructed by isoparametric transformation. The conformal distance field is generated by combining the lattice type and design parameters of the current hexahedral grid and the lattice information of the adjacent hexahedral grids. Step 4: Based on the axis-aligned bounding box of each hexahedral mesh, the search range is expanded to cover the adjacent boundary voxels, the triangular facets stored in the adjacent boundary voxels are extracted, and the closest distance value from the sampling point to the associated triangular facet within the conformal region of each hexahedral mesh unit is calculated to form the local distance field of the thin-walled model, and the inner wall distance field of the model is generated through implicit bias; Step 5: Perform implicit Boolean fusion on the lattice distance field and the inner wall distance field to generate a fused conformal distance field; Step 6: Use the marching cubes algorithm to extract the isosurface from the conformal distance field and output the STL model manufacturing file that integrates the dot matrix and the inner wall; Step 7: Merge the triangular facet information of the original thin-wall model, the lattice and the inner wall fused STL model manufacturing file to form a complete thin-wall lattice sandwich structure model manufacturing file.
2. The implicit generation method of a thin-walled lattice sandwich structure according to claim 1, characterized in that: The hexahedral meshing method in step 1 is to use the central axis algorithm to generate a hexahedral mesh that conforms to the geometry of the thin-walled model; and the hexahedral meshing thickness h2 satisfies the following relationship: h1 <h2<h0 Where h0 is the total thickness of the thin wall, and h1 is the actual thickness of the lattice filling area.
3. The implicit generation method of a thin-walled lattice sandwich structure according to claim 1, characterized in that: The specific steps of voxelization in step 2 include: S1, axis-aligned bounding box based on thin-wall model, dividing uniform voxel grid in 3D space; S2. For each triangular facet, filter the candidate voxels covered by its bounding box and use the separating axis theorem to detect the intersection between the voxel and the facet; S3. Mark the intersecting voxels as boundary voxels and store the corresponding triangle patch indexes. After all patches are processed, the complete model boundary voxel information is obtained.
4. The implicit generation method of a thin-walled lattice sandwich structure according to claim 1, characterized in that: The method for constructing the conformal lattice distance field in step 3 includes: S1. Mapping the mesh vertices of the hexahedral mesh unit from the local coordinate system of the lattice unit cell to the global coordinate system of the model to be filled by isoparametric transformation to construct a conformal distance field; S2. When generating a conformal lattice distance field based on the lattice type and design parameters of the current hexahedral mesh cell and the lattice information of the adjacent hexahedral mesh cells, the lattice parameters of up to 26 adjacent hexahedral mesh cells are incorporated to ensure geometric continuity at the boundary.
5. The implicit generation method of a thin-walled lattice sandwich structure according to claim 1, characterized in that: The method to expand the search range in step 4 is: S1. Based on the axis-aligned bounding box of the hexahedron, the expansion range along each coordinate axis is H = Δh + r node , get the search bounding box, where Δh = (h0-h1) / 2 is the inner wall offset thickness, r node is the maximum corner radius of the lattice node; S2. Extract boundary voxels and their corresponding triangles from the voxels covered by the expanded search bounding box.
6. The implicit generation method of a thin-walled lattice sandwich structure according to claim 1, characterized in that: The formula for implicit Boolean fusion in step 5 is: Among them, f lattice (x) is the lattice distance field, which includes the lattice parameters of the current hexahedral grid unit and its adjacent units; L is the number of lattice cells, including up to 26 adjacent cells of the current lattice cell; d(li,x) is the closest distance from the sampling point x in the current hexahedral grid conformal region to the surface of the lattice cell li; f bound (x) is the inner wall distance field, T is the set of boundary triangles obtained by searching the bounding box area, d(ti,x) is the closest distance from the sampling point x in the current hexahedral mesh conformal area to the surface of the boundary triangle ti; s(x) is the sign function, if x is -1 inside the model and 1 outside the model, it is determined by the pseudo-normal method; Δh is the inner wall offset thickness; the fusion process introduces the fusion fillet radius r blend To eliminate sharp edges.
7. The implicit generation method of a thin-walled lattice sandwich structure according to any one of claims 1 to 6, characterized in that: Steps 3 to 6 are executed in parallel in batches on the GPU, specifically including: S1. Divide the hexahedral mesh into preset batches based on the GPU hardware computing power, where each batch contains a preset number of hexahedral mesh units. S2. Allocate multiple GPU thread blocks to each hexahedral grid cell in each batch, and use threads in the multiple GPU thread blocks to parallely calculate distance field values for all sampling points in the corresponding hexahedral grid cell. S3. Execute steps 3 to 6 independently for each batch of hexahedral mesh elements and output them to the same manufacturing file; S4. After all batches are processed in a loop, they are finally merged to generate an STL model manufacturing file with a complete fusion of the dot matrix and the inner wall.