A method for additive manufacturing of a full-mouth implant-supported fixed denture framework

Abstract

This invention discloses an additive manufacturing method for full-mouth implant-supported fixed prostheses, belonging to the field of structural engineering technology. Addressing the complex geometric features of full-mouth implant-supported fixed prostheses, this invention utilizes a GPU-accelerated multi-objective optimization algorithm to optimize the placement angle, achieving minimal or no support in key areas such as the abutment interface. Simultaneously, it designs a truncated cubic lattice-column composite support structure, combined with targeted support designs for overhanging areas and edges, solving the problems of redundant support and insufficient molding accuracy in traditional methods. This significantly improves additive manufacturing accuracy and meets clinical needs.

Description

Technical Field

[0001] This invention relates to the field of oral medicine technology, and more specifically to an additive manufacturing method for a full mouth implant-supported fixed prosthesis framework. Background Technology

[0002] Currently, full-mouth implant-supported fixed dentures (ISFCD) are widely manufactured using laser powder bed fusion (LPBF) technology, which improves molding accuracy by optimizing placement angles and support structure design. Existing technologies typically employ multi-objective evolutionary algorithms (such as NSGA-II) to optimize framework placement angles, reducing overall overhang area and support requirements in critical areas (abutment interface, screw channels); simultaneously, commercial software (such as Magics) is used to generate columnar or block-shaped support structures to resist thermal deformation.

[0003] However, existing technologies have significant limitations: on the one hand, the general algorithms of traditional commercial additive manufacturing preprocessing software do not fully consider the unique geometric characteristics of ISFCD scaffolds—which include complex freeform surfaces such as dental arch curvature, as well as regular curved surfaces such as abutment interfaces and screw channels, and have significant individual differences, leading to the generation of unnecessary supports in key areas, affecting molding accuracy; on the other hand, the optimization of placement angles relies heavily on multi-objective optimization algorithms (such as NSGA-II), but the geometric model of ISFCD scaffolds is complex (containing hundreds of thousands of triangular facets), and traditional algorithms are computationally expensive and inefficient, making it difficult to quickly solve for the optimal angle; in addition, existing support structures (such as columnar, blocky, and tree-like structures) have problems such as difficulty in removal, powder locking, or insufficient resistance to deformation, and the applicability of lattice support topologies is not yet clear, resulting in LPBF-molded scaffolds often exceeding the clinically acceptable gap range due to deformation.

[0004] Therefore, how to propose an additive manufacturing method for full-mouth implant-supported fixed prosthesis frameworks to achieve efficient placement angle optimization and geometric feature adaptation of composite support generation, while ensuring clinical accuracy and improving manufacturing efficiency, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides an additive manufacturing method for full-mouth implant-supported fixed prosthesis frameworks, which can achieve efficient optimization of placement angle and adaptive geometric feature generation of composite support, thereby improving manufacturing efficiency while ensuring clinical accuracy.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for additive manufacturing of a full-mouth implant-supported fixed denture framework, comprising:

[0008] Reconstruct the topology model of the ISFCD stent;

[0009] Based on the aforementioned topological model, key geometric features of implant-supported fixed denture frameworks are identified, and feature patches are output.

[0010] Based on the topology model and the feature patches, an objective function is constructed, and the objective function is solved using a GPU-accelerated multi-objective optimization algorithm to obtain the optimal placement angle.

[0011] At the optimal placement angle, the area to be supported of the ISFCD bracket model and the contour projection of the area to be supported on the XOY plane are generated.

[0012] Grid nodes are generated at fixed intervals in the XOY plane. Valid grid nodes are determined by the intersection of the grid nodes with the contour of the area to be supported. For valid grid nodes, the intersection points with the facets of the support model are calculated along the Z-axis, and the intersection points within the area to be supported are taken as support points.

[0013] Based on the support points, a truncated cubic lattice support structure and a columnar support structure are generated, wherein the truncated cubic lattice support structure is generated above the effective grid cells, and the columnar support structure is generated in the region where there are effective nodes but no effective grid cells.

[0014] Identify the overhanging edge to be supported in the ISFCD support model and generate a plate-shaped support structure facing the overhanging edge to be supported;

[0015] By integrating the truncated cubic lattice support structure, columnar support structure, and plate-like support structure, the final support structure mesh is output, completing the design of the full mouth implant fixed denture framework.

[0016] Preferably, the topology model for reconstructing the ISFCD stent includes:

[0017] Read the STL file of the ISFCD support and extract the 3D coordinates of the vertices and the vertex indices of the face patches;

[0018] Create arrays of vertex, half-edge, and face structures, and unify geometric transformations into matrix operations using homogeneous coordinates;

[0019] Calculate the normal vector and area of ​​the face, locate the edge index through the half-edge hash table, and supplement the topological associations of vertices, edges, and faces.

[0020] Preferably, based on the topological model, key geometric features of implant-supported fixed denture frameworks are identified, and feature patches are output, including:

[0021] Select the seed facet corresponding to the base interface and screw channel;

[0022] The region growing method based on dihedral angles is used to identify feature regions. The growth rule is that the ratio of the dot product of the normal vectors of two adjacent facets to the product of the normal vector magnitudes is not less than the cosine value of the dihedral angle threshold.

[0023] The patches corresponding to the feature regions are marked as feature patches.

[0024] Preferably, constructing the objective function based on the topology model and the feature patches includes:

[0025] Determine whether a surface is a hanging surface based on the normal vector of the surface in the topological model;

[0026] Calculate the overall overhang area based on the overhanging surface. ;

[0027] Select the feature patches from the overhanging patches and calculate the overhang area of ​​key local regions. The objective function is constructed as follows:

[0028] ;

[0029] ;

[0030] In the formula, x Decision variables ,in This indicates the orientation angle of the part around the X-axis. Indicates the orientation angle of the part around the Y-axis; Let i be the i-th face, and area be the face. area, This is an indicator function that outputs 1 when the condition is true and 0 when the condition is false. surface panel It is a hanging surface. Surface plate For feature patches.

[0031] Preferably, the objective function is solved using a GPU-accelerated multi-objective optimization algorithm to obtain the optimal placement angle, including:

[0032] Initialize the particle swarm, with particle positions determined by rotation angles around the X and Y axes. Randomly generated within [0, 2π], velocity =0;

[0033] Calculate the fitness of each particle on the GPU device. Common parameters are stored in constant memory.

[0034] Update the external archive, filter non-dominated solutions based on Pareto dominance and adaptive grid partitioning, and maintain archive diversity using a roulette wheel strategy;

[0035] The particle velocity and position are iteratively updated based on inertia weight, cognitive factor and social factor, and a polynomial mutation operator is introduced to avoid local optima;

[0036] After k iterations, the solution with the smallest key local overhang area and the smallest overall overhang area is selected from the external archive as the optimal placement angle.

[0037] Preferably, at the optimal placement angle, generating the area to be supported of the ISFCD support model and the contour projection of the area to be supported on the XOY plane includes:

[0038] Select overhanging patches to form overhanging areas;

[0039] Calculate the minimum envelope circle of the overhanging region projected onto the XOY plane. If the diameter of the minimum envelope circle is greater than the maximum formable bridge length of the laser powder bed melting process... l If so, the suspended area is determined to be the area to be supported;

[0040] Extract half of the boundary of the area to be supported and generate a counterclockwise or clockwise polyline outline;

[0041] The vertices of the polyline are projected onto the XOY plane and merged into a two-dimensional projection of the region to be supported.

[0042] Preferably, a first grid is generated at fixed intervals in the XOY plane, and valid grid nodes are determined by the intersection of the grid nodes in the first grid with the contour of the region to be supported, including:

[0043] Based on the projected rectangular boundary of the AABB bounding box of the ISFCD bracket model optimized for the best placement angle in the XOY plane, a first mesh is generated at a fixed interval; the fixed interval is... ;

[0044] The influence range of a grid node in the first grid is defined as a rectangle centered on the grid node with a side length of 2L; where L is a preset influence diameter.

[0045] If the area of ​​influence intersects with the contour projection of the region to be supported in the XOY plane, then G is determined to be... i,j A valid grid node.

[0046] Preferably, the determination process for the effective grid cell is as follows:

[0047] Based on the projection rectangle boundary of the AABB bounding box of the ISFCD bracket model optimized for the best placement angle in the XOY plane, a second mesh is generated with the new spacing being an integer multiple of the support point spacing.

[0048] For any grid cell in the first grid, if more than half of the grid nodes in the corresponding first grid of the grid cell are valid grid nodes, then the grid cell is determined to be a valid grid cell.

[0049] Preferably, generating a truncated cubic lattice support structure and a columnar support structure based on the support points includes:

[0050] Calculate the z-coordinates of the first intersection points of the k effective mesh nodes corresponding to the effective mesh element with the ISFCD support model along the Z-axis direction, which are respectively ;

[0051] The number of lattice units above the effective grid unit is determined based on the z-coordinate, using the following formula:

[0052] ;

[0053] In the formula, To support the height of contact, The minimum angle for connecting the branches; , These are the side lengths of the first grid in the X and Y directions, respectively;

[0054] The cross-sectional cubic lattice support structure body is generated according to the number of lattice units, and corresponding support contacts and support connections are generated.

[0055] In regions with non-effective mesh cells but effective nodes, for any effective mesh node, if the number of intersections with the ISFCD support model along the z-axis is ≥2 and there exists a natural number k such that the 2k+2th intersection is located within the area to be supported, then a columnar support with a cross-sectional dimension equal to the support contact width is generated between the 2k+1th and 2k+2th intersections; wherein the intersections are arranged in ascending order of z-coordinate.

[0056] Preferably, identifying the overhanging edge to be supported in the ISFCD support model and generating a plate-shaped support structure facing the overhanging edge to be supported includes:

[0057] The overhanging edge to be supported simultaneously meets the following judgment conditions:

[0058] (1) The sum of the Z coordinates of two vertices of an edge is less than the Z coordinate of the other vertex of the same face.

[0059] (2) The Z-coordinate of the sum of the normal vectors of the two faces containing the edge is negative;

[0060] (3) Neither of the two faces containing the edge is a hanging face;

[0061] Traverse all half-sides and determine whether the half-side is a hanging half-side according to the judgment condition; connect the hanging half-sides into multiple line segments and filter out the multiple line segments that do not need support.

[0062] The filtered polyline vertices and their projected vertices onto the XOY plane are connected to form a curved surface, and the curved surface is shelled into a solid plate-like support.

[0063] As can be seen from the above technical solution, the present invention discloses a method for additive manufacturing of a full-mouth implant-supported fixed denture framework, which has the following advantages compared with the prior art:

[0064] 1. A GPU-accelerated placement angle optimization algorithm is proposed to achieve fewer or even no support structures in key local areas (abutment interface and screw channel) of full-mouth implant fixed denture framework;

[0065] 2. An algorithm for generating a lattice-column composite support structure based on a high-strength and lightweight truncated cubic lattice structure is proposed. This composite support structure can ensure the forming accuracy of the full-mouth implant fixed denture framework. Attached Figure Description

[0066] Figure 1 A flowchart of the design method provided for this invention;

[0067] Figure 2 To reconstruct the topology of the STL mesh;

[0068] Figure 3(a) shows the initial seed patch of the manually selected feature region; Figure 3(b) shows a single geometric feature automatically identified; Figure 3(c) shows all the identified geometric features.

[0069] Figure 4 This is the external archive after 100 iterations;

[0070] Figure 5(a) shows the suspended area of ​​the full-mouth implant fixed denture framework; Figure 5(b) shows the area to be supported; Figure 5(c) shows the projection of the outline of the area to be supported onto the XOY plane;

[0071] Figure 6(a) shows the generation of mesh nodes for generating support points; Figure 6(b) shows the generation of support points; Figure 6(c) shows the generation of mesh nodes for generating lattice support structures.

[0072] Figure 7 A schematic diagram of the lattice support structure above an effective grid cell;

[0073] Figure 8(a) shows the lattice support structure; Figure 8(b) shows the columnar support structure; Figure 8(c) shows the composite support structure formed by combining the two.

[0074] Figure 9(a) shows a suspended polyline; Figure 9(b) shows a plate-like support structure. Detailed Implementation

[0075] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0076] This invention discloses an additive manufacturing method for a full-mouth implant-supported fixed denture framework, with reference to... Figure 1 This includes the following steps:

[0077] S1. Reconstruct the topological model of the ISFCD scaffold, such as Figure 2 As shown.

[0078] The STL file is read to extract the 3D coordinates of each vertex (geometric model) and the indices of the vertices contained in each facet (topological model). However, simply knowing the indices of the vertices contained in each facet is insufficient; more topological information is needed. Therefore, a new model is created to describe the STL mesh. To facilitate the subsequent unification of geometric transformations (rotation and translation) into matrix multiplication, homogeneous coordinates are used respectively. , Representing vertices (Point) and vectors (Vector), the geometric vertices of the model are represented by the Point structure array `pointArray`. A Vertex structure array `vertexArray`, a HalfEdge structure array `edgeArray`, and a Facet structure array `facetArray` are created. The member variables of the Vertex structure contain the index `pointIndex` of the corresponding geometric vertex in the `pointArray`. The member variables of the HalfEdge structure include the vertex index `vertexIndex` and the index of its opposite edge `OppositeEdgeIndex`. The vertex index `vertexIndex` is equal to the index `edgeIndex` of the half edge radiating from that vertex. The quotient obtained by dividing the vertex index `vertexIndex` or the face index `edgeIndex` by the integer 3 is the facet index `facetIndex` containing that vertex or half edge. The member variables of the Facet structure contain the face's normal vector `normal`, area `area`, and a Boolean value `isFeature` indicating whether it is a feature face. (The last sentence, "Let the face...", is a separate, unrelated statement and can be omitted.) The three vertices The coordinates are respectively , , Calculate the normal vector of the face using the following formula. : ,Right now

[0079] ;

[0080] Then calculate the unit normal vector of the patch. Save the data and calculate the area of ​​the patch. And save it, among which This allows for a quick and accurate understanding of the relationships between points, edges, and faces, excluding the opposite edge of an edge and the edge containing a vertex. A half-edge hash table is constructed to quickly locate the opposite edge index of an edge. If the hash codes of multiple edges map to the same integer, these vertices are arranged into a linked list. The key of each linked list node is the index of the edge projected from that vertex, i.e., the vertex index. Then, an array of sequential lists for the edges containing a point is created. The elements in the array are the sequential lists corresponding to each point, and the elements in the sequential lists are the edge indices (edgeIndex) of the edges containing that point. All edges are set to unused. The model faces are traversed. For each of the three vertices of a face, if the edge containing a certain vertex is unused, its index (edgeIndex) is added to the sequential list of the point corresponding to that vertex, and the edge and its opposite edge are marked as used. After the traversal is complete, all edges are restored to unused status.

[0081] S2. Identify key geometric features of implant-supported fixed denture frameworks based on topological models and output feature patches.

[0082] After creating the topology model, the key geometric features of the implant-supported fixed prosthesis framework—the abutment interface and screw channels—were identified. This area is a regular-shaped surface, while other areas of the bridge are free-form surfaces. For the distinction between regular and free-form surfaces, the dihedral angle is a robust and practical geometric feature. Seed faces were manually selected using the VTK library, and the feature regions containing these seed faces were identified using region growing. The growing rule was that the included dihedral angle between two adjacent faces should not exceed a certain threshold. Let the normal vectors of these two faces be denoted. and Then it needs to satisfy ,Right now

[0083] ;

[0084] In this embodiment, a structure containing a std::vector type variable vector and an integer variable front is used to represent the feature region (the initial value of front is 0). After selecting several seed patches, all the key geometric feature regions of the ISFCD scaffold can be identified (Figure 3). According to the feature region queue Q, the corresponding patches in the topology model patch array are marked as feature patches.

[0085] S3. Construct an objective function based on the topology model and feature patches, and solve the objective function using a GPU-accelerated multi-objective optimization algorithm to obtain the optimal placement angle.

[0086] In this embodiment, the following two objective functions or fitness values ​​are selected for optimization: (1) Overall overhang area (2) Overhang area of ​​key local areas All units are mm 2 Decision variables ,in This indicates the orientation angle of the part around the X-axis. The angle of placement of the part around the Y-axis is expressed in radians. The optimization problem involved in this embodiment can be described by the following mathematical model:

[0087]

[0088] Iterate through the facetArray array of facet structures and determine the angle of rotation of the model around the X-axis. The angle of rotation around the Y-axis is Then, it becomes a suspended surface. The method is as follows: After the model is rotated, the surface... normal vector Compared with the original normal vector The relationship is as follows:

[0089] .in, , Given the rotation matrices around the X-axis and Y-axis, respectively, the normal vector coordinates of the rotated face can be calculated:

[0090] ;

[0091] Because the original normal vector of the surface For unit vectors, rotation matrix Since it is an orthogonal matrix, the normal vector of the rotated facet is... It remains a unit vector. By definition, if the surface... normal vector Unit vector in the negative direction of the Z-axis The included angle is less than the maximum formable angle. That is, satisfying the following formula:

[0092] ;

[0093] Such a facet is considered an overhang facet and is denoted as isOverhang. After traversing all faces of the model, the objective function can be calculated:

[0094] .

[0095] Furthermore, the facetArray of the topology model is copied from host memory to the device's global memory. An array of Particle structures is used to represent the facet array. n A population S of 1 particle. Particle P. i Has speed ,Location Adaptability Optimal fitness Optimal position The attribute is t, where t represents the iteration number of the population. The population S is copied from host memory to the device's global memory. Initially, particle P... i speed Set the value to 0 to calculate the particle's position. Components in the j-th dimension:

[0096] ;

[0097] Where j is 1 or 2, To obey random numbers, , Calculate the fitness of the particles. Then calculate the optimal position of the particle. Optimal fitness The particle swarm is initialized using a kernel function, with one particle initialized per thread, allowing simultaneous initialization of all particles. Temporary variables generated during initialization are stored in registers. The cosine value of the angle threshold... Variables that are applicable to all particles are stored in constant memory that can be read by all threads.

[0098] An external archive A is represented using a similar data structure, where the number of particles in A is specified to not exceed [a certain value]. n 2. Since updating A requires comparison between particles, and communication between threads in different thread bundles of the kernel function and device function is inconvenient, population S is copied from the device to the host after each update. Particles from S are added to A, and the Pareto dominance relationship between particle positions in A is determined; only particles with dominant positions are retained. If the number of particles in A exceeds m, a particle is sequentially deleted until the number of particles in A reaches m. n 2. The method for selecting particles each time is as follows: adaptively divide the target space into grids based on fitness. Calculate the particle P in A at the t-th iteration. iThe grid cell number is used. A roulette wheel selection method is employed to select the grid cell. To ensure population diversity, grid cells with larger crowding distances (i.e., a larger number of particles) have a higher probability of being selected. A particle is randomly selected from the chosen grid cell. After updating A, adaptive grid partitioning and the roulette wheel selection method are still used to select a leader particle from A, and its position becomes particle P in population S. i The corresponding global optimal solution To ensure population diversity, grid cells with smaller crowding distances are more likely to be selected.

[0099] After selecting the globally optimal solution set, it is copied from the host memory to the device's global memory. The kernel function is used to update each particle in the population S. Particle P is calculated according to the following formula. i speed at the (t+1)th iteration :

[0100] ;

[0101] Where c1 is the cognitive learning factor and c2 is the social learning factor. and To obey random numbers, , These represent the particle's position in the previous iteration and its optimal position, respectively, along with the inertia weight. The iteration count decreases linearly with the number of iterations t. If the particle's velocity is too high, i.e. ,in, To be constant, the particle velocity needs to be limited:

[0102] ;

[0103] Then the position of the particle is calculated:

[0104] ;

[0105] To increase particle diversity and avoid getting trapped in local optima, a polynomial mutation operator is used to adjust particle positions. The probability of each particle's position undergoing a mutation is... P m The iteration count decreases linearly with the number of iterations t. If the particle's position changes abruptly, the abrupt change value of the particle's position is calculated. Particle position The component in the j-th dimension is:

[0106] ;

[0107] If the particle position If the boundary is exceeded in the j-th dimension, the handling method is as follows:

[0108] ;

[0109] in, This is a local minimum. The fitness at iteration t+1 can then be calculated. In the (t+1)th iteration, if Pareto Domination Then the optimal position of the particle ;if Pareto Domination , ;otherwise, exist and Choose either one. After updating all particles in population S, copy them from the device to the host. Update the external archive A on the host. Iterate a certain number of times k. Figure 4 This is the external archive after 100 iterations.

[0110] A particle is selected from the external archive A after k iterations. The position of this particle, representing the global optimal solution, is used as the optimal placement angle for the model. The method for selecting the particle is as follows: within the two-dimensional space of the overhang area of ​​the critical local region versus the overhang area of ​​the global region, the particle in the lower left region is selected, i.e., the particle with the smallest overhang area in both the critical local region and the global region; to reduce deformation, the height difference in the Z direction of the part at the corresponding placement angle is selected from these regions. z max - z min The smallest particle. After selecting the optimal particle, calculate the normal vector coordinates of each facet of the model after rotating it according to the particle's position. Similarly, the vertex coordinates of the rotated facets can be calculated, thus obtaining the model rotated according to the particle's position. Let the minimum and maximum values ​​of the vertex coordinates of all vertices in the Z direction of the ISFCD support model at the optimal placement angle be respectively... z min , z max To ensure the printed model can be easily cut from the substrate, the lowest point of the model is maintained at a certain height h from the substrate along the Z-axis. For any point... The corresponding new point after translation The Z coordinate is .

[0111] The geometric features related to support generation include overhanging patches, overhanging edges, and overhanging points. For complex freeform surface structures such as dentures, the patch size is usually small. The geometric features related to isolated overhanging points of such parts are also small and do not appear in critical local areas such as the support interface, so it is unnecessary to use these overhanging points as support points. Therefore, in this embodiment, overhanging points are not considered; only overhanging patches and overhanging edges are considered.

[0112] S4. At the optimal placement angle, generate the area to be supported of the ISFCD support model and the contour projection of the area to be supported on the XOY plane.

[0113] Iterate through all faces of the model, find overhanging faces, and store their indices in the overhanging face set. A vector is used to represent the set of overhanging regions. The elements in the vector represent the overhanging region C. According to... Generate model overhang region The generated overhanging regions are shown in Figure 5(a). Due to the self-supporting effect of the model parts, not all overhanging regions require additional support structures. Calculate the minimum envelope circle of the overhanging region projected onto the XOY plane. If the minimum envelope circle diameter d and the maximum formable bridge length l of LPBF satisfy... If the overhanging region C is considered to be the region to be supported, then the generated region to be supported is shown in Figure 5(b). Next, the projection of the boundary contour of the region to be supported onto the XOY plane is calculated. The boundary contour of the region is represented using a polyline vector. A double-ended queue (polyline) is used to represent the polyline, with its elements being the vertex indices (pointIndex) of the polyline. The contour of the region is generated based on the edgeVector, the half-edge vectors that make up the boundary of the region U to be supported. If the closed polylines generated according to the above method are located on the outer boundary of the region, their order is counter-clockwise; if they are located on the inner boundary of the region, their order is clockwise. When no unused half-edge is found in the edgeVector array, the boundary contours of all regions to be supported are obtained. Then, the vertex indices of the polylines are replaced with the coordinates of the corresponding vertices, and they are projected onto the XOY plane. The projections of the boundary contours of each region are merged, and the merged boundary contour is... That is, the two-dimensional projection of the area to be supported by the model. As shown in Figure 5(c).

[0114] S5. Generate mesh nodes at fixed intervals in the XOY plane. Determine the valid mesh nodes by the intersection of the mesh nodes with the contour of the area to be supported. For valid mesh nodes, calculate the intersection points with the facets of the support model along the Z-axis direction, and take the intersection points in the area to be supported as support points.

[0115] Let the rectangle of the AABB bounding box of the model in the XOY plane after rotation and translation have the following boundaries in the X and Y directions: , , , The rectangle is meshed, with each mesh cell being a square with a side length of 'a', resulting in the first mesh. Since the maximum formable bridge length 'l' of LPBF-printed titanium alloy bridges is possible, to ensure that all areas of the envelope rectangle are covered by the influence range of the nodes, the side length of the mesh cells is preset. To ensure that all subsequent mesh elements and lattice elements precisely cover the entire envelope rectangle, it is necessary to fine-tune the side lengths of the mesh elements. Let the side lengths of the fine-tuned mesh in the X and Y directions be respectively... , The X and Y coordinates of the mesh nodes are stored using `std::vector`. The `IntersectPoint` structure records the coordinates of the intersection points and whether the face containing the intersection point belongs to the region to be supported. The `std::vector` array stores the intersection points of valid grid nodes with the model's vertical lines, where the first and second indices correspond to the X and Y coordinate indices of the grid nodes, respectively. The third dimension of the array stores the relevant intersection points. , where n is a natural number, and the number of intersection points must be even. Grid nodes The area of ​​influence is a circle with a diameter of L, which can be simplified as a circle consisting of mesh nodes. Centered on, with side length as rectangle .if and Intersection, that is Then the grid nodes are considered to be For effective mesh nodes, 2D polygons X and Y coordinates of the centroid Treat as a grid node The X and Y coordinates of the relevant support points are projected. This allows us to identify all valid grid nodes and update their X and Y coordinates. The grid nodes of the generated first grid are shown in Figure 6(a).

[0116] The model iterates through all the faces it contains, calculating the 2D AABB bounding box projected onto the XOY plane for each face. The intersection points of the mesh nodes within this bounding box are then selected; the vertical lines containing these intersection points may intersect with the face. Further determination is needed to verify whether the vertical lines containing the intersection points actually intersect the face. (Intersection points) The necessary and sufficient condition for the intersection of the relevant vertical line and the face is that the point... The triangle projected onto the XOY plane from this surface Inside. This allows for the selection of pieces that match the surface. Intersecting grid nodes Projection of relevant support points. Then, calculation of mesh nodes. The Z-coordinate of the intersection point of the vertical line where the projection of the relevant support point lies and the surface is: ,in, For dough The normal vector is obtained. This allows us to calculate the coordinates of the intersection points between the model and the vertical lines containing all mesh nodes, and to mark whether each intersection point belongs to the area to be supported in the model. The intersection points of the vertical lines projected onto the support points of the same mesh node are sorted in ascending order of Z-coordinate. The generated support points are shown in Figure 6(b).

[0117] S6. Generate truncated cubic lattice support structures and columnar support structures based on support points, wherein the truncated cubic lattice support structure is generated above the effective grid cells, and the columnar support structure is generated in the region where there are effective nodes but no effective grid cells.

[0118] The process for determining valid mesh cells is as follows:

[0119] Based on the projection rectangle boundary of the AABB bounding box of the ISFCD bracket model optimized for the best placement angle in the XOY plane, a second mesh is generated with the new spacing being an integer multiple of the support point spacing.

[0120] For any grid cell in the first grid, if more than half of the corresponding grid nodes in the first grid are valid grid nodes, then the grid cell is determined to be a valid grid cell.

[0121] In this embodiment, to ensure that the forces on the part at each support point in the X and Y directions are as consistent as possible, the size of the lattice support structure unit is set to an integer multiple of the support point spacing. If the truss lattice support structure unit is set to be equal to the support point spacing, considering the width of the single melt channel of LPBF molding titanium alloy is approximately 0.15 mm, powder adhesion issues, and the need for the powder to be able to be poured out of the support structure after printing, the adjustable range of the truss lattice structure's column size would be very narrow. Therefore, in this invention, the size of the truss lattice structure unit is set to twice the support point spacing. Let the side lengths of the lattice unit along the X, Y, and Z axes be respectively... , and The resulting mesh, obtained by sampling the grid again on the XOY plane of the AABB bounding box of the part model, is called the second mesh. Grid The element lengths in the X and Y axes are respectively and For grids any unit Its corresponding grid GThe four units, with their four nodes Corresponding to the grid G 9 nodes If the grid G If more than half of these 9 nodes are valid, then the unit is considered valid. "Valid". The generated mesh nodes for the lattice support structure are shown in Figure 6(c).

[0122] This embodiment uses high-strength, lightweight truncated cubic lattice units as the basic units of the lattice support structure. If the unit "Valid," generating a truncated cubic lattice support structure above this unit. This support structure comprises the main lattice support structure and support contacts. Assume that among these 9 mesh nodes, k nodes have vertical lines that intersect with the part model. Let the Z-coordinates of the first intersection points of these vertical lines with the part model along the Z-axis be... Then the unit The number of upper lattice units is .in, To support the height of contact, The minimum angle for connecting the branches. In the unit... After the main body of the cross-sectional cubic lattice support structure is generated above, the corresponding support contacts and support connections are generated. These 9 nodes can be divided into three categories: corner points. Edge Point and center point The support contact and connection below the corner points, edge points, and center points are as follows: Figure 7 The first three are shown, and the last one is the merged lattice support structure. The generated lattice support structure is shown in Figure 8(a). If the unit Not valid, but grid G At least one of the nine nodes is valid. To control the volume of the support structure, a columnar support structure is generated for the valid mesh nodes. For any mesh node... If the vertical line it is located at intersects with the part model at more than two points, and the intersection points are... If the region to be supported in the model is located (k is a natural number), then from the intersection point... to the intersection A columnar support structure with a cross-sectional dimension equal to the support contact width is generated between the two structures. The generated columnar support structure is shown in Figure 8(b). The composite support structure formed by merging the lattice support structure and the columnar support structure is shown in Figure 8(c).

[0123] S7. Identify the overhanging edge to be supported in the ISFCD support model and generate a plate-shaped support structure facing the overhanging edge to be supported.

[0124] The definition of a dangling edge is as follows: (1) the two vertices of the edge and The Z-coordinate of the edge is less than the other vertex of the face containing that edge. and The Z-coordinate of ), that is:

[0125] ;

[0126] (2) The normal vectors of the two faces containing the edge n 1 and n The Z-coordinate of the sum of 2 is negative, that is:

[0127] ;

[0128] However, not all overhanging edges require additional support structures. If one of the two faces containing the overhanging edge is an overhanging face, then the support structure designed earlier for the overhanging face in this embodiment can function. Therefore, we only need to consider overhanging edges where neither of the corresponding faces is an overhanging face, i.e.:

[0129] ;

[0130] Traverse all halves of the model and determine if each half is a suspended edge that has not been used. If the condition is met, mark the half as suspended. Similarly, connect the suspended halves into polylines. The suspended polylines of the ISFCD support are shown in Figure 9(a). Not all suspended polylines require support. The method for filtering suspended polylines is the same as for filtering suspended areas. Polylines to be supported may not exist.

[0131] Since it is impossible to determine in advance whether there are still polylines to be supported after filtering, this embodiment designs a plate-like support structure for polylines. Regardless of whether the polyline is closed, the vertices of the polyline are projected onto the XOY plane, and the vertices of the polyline and their projected vertices onto the XOY plane are connected to form a curved surface. This curved surface is then shelled into a solid plate-like support with a certain thickness.

[0132] S8. Integrate the truncated cubic lattice support structure, columnar support structure and plate support structure to output the final support structure mesh, and complete the design of the full mouth implant fixed denture framework.

[0133] If a support structure facing the overhang exists (as shown in Figure 9(b)), merge the support structures facing the overhang and the overhang to obtain the final support structure mesh, and output it in STL format.

[0134] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0135] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for additive manufacturing of a full-mouth implant-supported fixed denture framework, characterized in that, include: Reconstruct the topology model of the ISFCD stent; Based on the aforementioned topological model, key geometric features of implant-supported fixed denture frameworks are identified, and feature patches are output, including: Select the seed facet corresponding to the base interface and screw channel; The region growing method based on dihedral angles is used to identify feature regions. The growth rule is that the ratio of the dot product of the normal vectors of two adjacent facets to the product of the normal vector magnitudes is not less than the cosine value of the dihedral angle threshold. The patch corresponding to the feature region is marked as a feature patch; Constructing an objective function based on the topological model and the feature patches includes: Determine whether a surface is a hanging surface based on the normal vector of the surface in the topological model; Calculate the overall overhang area based on the overhanging surface. ; Select the feature patches from the overhanging patches and calculate the overhang area of ​​key local regions. The objective function is constructed as follows: ； ； In the formula, x Decision variables ,in This indicates the orientation angle of the part around the X-axis. This indicates the placement angle of the part around the Y-axis; N is the total number of facets. Let i be the i-th face, and area be the face. area, This is an indicator function that outputs 1 when the condition is true and 0 when the condition is false. Surface plate It is a hanging surface. Surface plate For feature patches; The objective function is solved using a GPU-accelerated multi-objective optimization algorithm to obtain the optimal placement angle, including: Initialize the particle swarm, with particle positions determined by rotation angles around the X and Y axes. Randomly generated within [0, 2π], velocity =0; Calculate the fitness of each particle on the GPU device. Common parameters are stored in constant memory. Update the external archive, filter non-dominated solutions based on Pareto dominance and adaptive grid partitioning, and maintain archive diversity using a roulette wheel strategy; The particle velocity and position are iteratively updated based on inertia weight, cognitive factor and social factor, and a polynomial mutation operator is introduced to avoid local optima; After k iterations, the solution with the smallest key local overhang area and the smallest overall overhang area is selected from the external archive as the optimal placement angle. At the optimal placement angle, the area to be supported of the ISFCD bracket model and the contour projection of the area to be supported on the XOY plane are generated. Grid nodes are generated at fixed intervals in the XOY plane. Valid grid nodes are determined by the intersection of the grid nodes with the contour of the area to be supported. For valid grid nodes, the intersection points with the facets of the support model are calculated along the Z-axis, and the intersection points within the area to be supported are taken as support points. Based on the support points, a truncated cubic lattice support structure and a columnar support structure are generated, wherein the truncated cubic lattice support structure is generated above the effective grid cells, and the columnar support structure is generated in the region where there are effective nodes but no effective grid cells. Identify the overhanging edge to be supported in the ISFCD support model and generate a plate-shaped support structure facing the overhanging edge to be supported; By integrating the truncated cubic lattice support structure, columnar support structure, and plate-like support structure, the final support structure mesh is output, completing the design of the full mouth implant fixed denture framework.

2. The additive manufacturing method for a full-mouth implant-supported fixed denture framework according to claim 1, characterized in that, Reconstructing the topology model of the ISFCD scaffold includes: Read the STL file of the ISFCD support and extract the 3D coordinates of the vertices and the vertex indices of the face patches; Create arrays of vertex, half-edge, and face structures, and unify geometric transformations into matrix operations using homogeneous coordinates; Calculate the normal vector and area of ​​the face, locate the edge index through the half-edge hash table, and supplement the topological association between vertices, edges, and faces.

3. The additive manufacturing method for a full-mouth implant-supported fixed denture framework according to claim 1, characterized in that, At the optimal placement angle, the region to be supported of the ISFCD support model and the contour projection of the region to be supported onto the XOY plane are generated, including: Select overhanging patches to form overhanging areas; Calculate the minimum envelope circle of the overhanging region projected onto the XOY plane. If the diameter of the minimum envelope circle is greater than the maximum formable bridge length of the laser powder bed melting process... l If so, the suspended area is determined to be the area to be supported; Extract half of the boundary of the area to be supported and generate a counterclockwise or clockwise polyline outline; The vertices of the polyline are projected onto the XOY plane and merged into a two-dimensional projection of the region to be supported.

4. The additive manufacturing method for a full-mouth implant-supported fixed denture framework according to claim 3, characterized in that, A first grid is generated in the XOY plane at fixed intervals. Valid grid nodes are determined by the intersection of the grid nodes in the first grid with the contour of the region to be supported, including: Based on the projected rectangular boundary of the AABB bounding box of the ISFCD bracket model optimized for the best placement angle in the XOY plane, a first mesh is generated at a fixed interval; the fixed interval is... ; The influence range of a grid node in the first grid is defined as a rectangle centered on the grid node with a side length of 2L; where L is a preset influence diameter. If the area of ​​influence intersects with the contour projection of the region to be supported in the XOY plane, then G is determined to be... i,j For a valid grid node, G i,j It is the grid node in the i-th row and j-th column of the first grid.

5. The additive manufacturing method for a full-mouth implant-supported fixed denture framework according to claim 4, characterized in that, The process for determining the effective grid cell is as follows: Based on the projection rectangle boundary of the AABB bounding box of the ISFCD bracket model optimized for the best placement angle in the XOY plane, a second mesh is generated with the new spacing being an integer multiple of the support point spacing. For any grid cell in the first grid, if more than half of the grid nodes in the corresponding first grid of the grid cell are valid grid nodes, then the grid cell is determined to be a valid grid cell.

6. The additive manufacturing method for a full-mouth implant-supported fixed denture framework according to claim 5, characterized in that, Based on the aforementioned support points, a truncated cubic lattice support structure and a columnar support structure are generated, including: Calculate the z-coordinates of the first intersection points of the k effective mesh nodes corresponding to the effective mesh element with the ISFCD support model along the Z-axis direction, which are respectively ; The number of lattice units above the effective grid unit is determined based on the z-coordinate, using the following formula: ； In the formula, To support the height of contact, The minimum angle for connecting the branches; , These are the side lengths of the first grid in the X and Y directions, respectively. For fixed spacing; The cross-sectional cubic lattice support structure body is generated according to the number of lattice units, and corresponding support contacts and support connections are generated. In regions with non-effective mesh cells but effective nodes, for any effective mesh node, if the number of intersections with the ISFCD support model along the z-axis is ≥2 and there exists a natural number k such that the 2k+2th intersection is located within the area to be supported, then a columnar support with a cross-sectional dimension equal to the support contact width is generated between the 2k+1th and 2k+2th intersections; wherein the intersections are arranged in ascending order of z-coordinate.

7. The additive manufacturing method for a full-mouth implant-supported fixed denture framework according to claim 1, characterized in that, Identify the overhanging edge to be supported in the ISFCD support model, and generate a plate-shaped support structure facing the overhanging edge to be supported, including: The overhanging edge to be supported simultaneously meets the following judgment conditions: (1) The sum of the Z coordinates of two vertices of an edge is less than the Z coordinate of the other vertex of the same face. (2) The Z-coordinate of the sum of the normal vectors of the two faces containing the edge is negative; (3) Neither of the two faces containing the edge is a hanging face; Traverse all half-sides and determine whether the half-side is a hanging half-side according to the judgment condition; connect the hanging half-sides into multiple line segments and filter out the multiple line segments that do not need support. The filtered polyline vertices and their projected vertices onto the XOY plane are connected to form a curved surface, and the curved surface is shelled into a solid plate-like support.

Images