Ordered inhibition method and device for transition features of complex model

Abstract

The invention relates to a method and a device for orderly inhibiting transition features of a complex model, and the method comprises the following steps: firstly constructing an attribute adjacency graph based on a boundary representation model, and recognizing a transition feature chain through dynamic sampling and principal curvature analysis; secondly, decomposing complex intersection features into ordered suppression queues by adopting a minimum leaf sub-graph preferential sorting strategy; in the inhibition stage, the topological structure of the model is directly modified by adopting an Euler operator, and geometric stitching is realized through intersection of a bottom layer curved surface; and finally, an incremental local updating mechanism is adopted, and only graph nodes and edges influenced by topological operation are corrected. According to the method, the problems of easy generation of self-intersection and low calculation efficiency in free-form surface model simplification are solved, and the efficiency and robustness of large-scale complex model simplification are remarkably improved while the topological effectiveness of the model is ensured.

Description

Technical Field

[0001] This invention relates to the field of preprocessing technology in computer-aided design (CAD), computer graphics, and computer-aided engineering (CAE), specifically to a method and apparatus for orderly suppression of transition features in complex models. Background Technology

[0002] In modern industrial product design, to meet aerodynamic performance, aesthetic requirements, manufacturing processes (such as casting draft and stress relief), and ergonomic needs, designers extensively use transition features in part models, primarily fillets and chamfers. These features are essential in the CAD modeling stage.

[0003] However, in downstream applications of products, especially in feature identification for finite element analysis (FEA) and computer-aided manufacturing (CAM), these fragmented transitional features often become obstacles. Transitional features typically have small radii of curvature and elongated geometries, which forces mesh generation algorithms to generate extremely fine meshes in these regions, leading to a surge in the number of elements, an exponential increase in computational costs, and even poor mesh quality (such as excessively large aspect ratios), resulting in non-convergence of the solution.

[0004] In CAM, to identify machining features such as cavities and hole systems, it is necessary to extract the macroscopic topological structure of the model. The presence of transitional features disrupts the direct adjacency relationships between surfaces, increasing the complexity of feature recognition algorithms.

[0005] Therefore, model simplification, especially the automatic suppression of transitional features, has become a key technology for CAD / CAE integration.

[0006] In existing technologies, transition feature suppression mainly falls into the following categories:

[0007] Design history-based suppression: This method disables or removes feature steps that generate rounded corners by traversing the feature tree. It is the most reliable method, but heavily relies on the native CAD format. In collaborative chains such as aerospace and automotive, models are often transmitted in intermediate formats like STEP and IGES, resulting in the loss of design history and rendering this method ineffective.

[0008] Voxel / pixel-based simplification: The model is converted into voxels, morphologically eroded and dilated, and then reconstructed. This method suffers from significant accuracy loss, cannot maintain the precise geometric boundaries of B-Rep, and is unsuitable for high-precision simulations.

[0009] Direct Modeling based on face extension and trimming: This is the current mainstream approach to "removing rounded corners". The principle is: delete the rounded corner face, then extend the adjacent faces (support faces) on both sides tangentially until they intersect, and then trim off the excess parts.

[0010] Defect Analysis: This method performs reasonably well when dealing with regular curved surfaces (planes, cylinders), but faces significant challenges when dealing with free-form surfaces (NURBS surfaces). The extension of a free-form surface is an ill-posed problem, and the curvature change of the extended portion is difficult to control.

[0011] Problem 1: The extended surfaces may self-intersect or diverge, making it impossible to find the intersection line.

[0012] Question 2: At complex Y-shaped or T-shaped rounded corner intersections, multiple extended surfaces are difficult to form a closed clipping boundary, which can easily lead to topological vulnerabilities (water-tightness issue).

[0013] Problem 3: This method relies too heavily on the stability of geometric calculations and has poor robustness.

[0014] Graph-based recognition and suppression: Feature recognition is achieved using Attribute Adjacency Graphs (AAGs). However, in the suppression phase, existing algorithms typically employ a "global reconstruction" strategy, meaning that for each feature deleted, the entire model is rescanned to reconstruct the AAG. For complex models containing thousands of faces, this approach becomes problematic. Operations at this level result in extremely low efficiency. Furthermore, a lack of a proper "deletion order" can disrupt the dependencies between external features when processing internal features, leading to topology collapse. Summary of the Invention

[0015] To address the problems existing in the prior art, the purpose of this invention is to provide a method and apparatus for orderly suppression of transition features in complex models. By modifying the topological logic and precisely stitching together the underlying geometry, a highly robust and efficient simplification of model features is achieved.

[0016] To achieve the above objectives, the present invention adopts the following technical solution:

[0017] A method for orderly suppression of transition features in complex models includes the following steps:

[0018] Step S1: Parse the input 3D B-Rep model, extract point, edge, and face topological elements to construct an attribute adjacency graph (AAG), perform mesh sampling and differential geometric analysis on the surface, calculate the mean curvature and Gaussian curvature, identify the edge-blend face (EBF) and vertex-blend face (VBF) in the model, and map these geometric attributes to the attribute labels of AAG nodes;

[0019] Step S2: Based on the edge attributes (spring edge, cross edge, termination edge) in the attribute adjacency graph, cluster adjacent transition surfaces into transition feature chains;

[0020] Step S3: Analyze the external connectivity and nesting relationship of each transition feature chain in the attribute adjacency graph, identify the minimum leaf subgraph, that is, the feature chain with weak connection to other feature groups, few faces and located at the topological edge, and construct a priority suppression queue based on the minimum leaf subgraph; where, the external connectivity is the number of edges in the transition feature chain that are connected to surfaces other than the chain.

[0021] Step S4: Based on Euler operators, perform topological suppression and geometric stitching on the current transition feature chain according to the priority suppression queue: generate the corresponding Euler operator sequence (KEF, KEV, KFMV) according to the feature type, delete the transition surface and modify the topological connection relationship so that the support surfaces that were originally separated by the transition surface are topologically adjacent; then calculate the intersection line of the bottom geometric surface of the support surface, and update the geometric information of the edges and vertices to close the model;

[0022] Step S5: After the suppression operation is completed, lock the affected topology region, delete the failed nodes and update the adjacent edges in the attribute adjacency graph, and perform incremental local update of the attribute adjacency graph.

[0023] Step S6: Iterate through steps S3 to S5 until all target transition feature chains are suppressed.

[0024] Furthermore, the transition surface identification method in step S1 specifically includes:

[0025] For any surface, in the parameter domain Internal partitioning The sampling grid is defined; the maximum principal curvature at each sampling point is calculated. and minimum principal curvature If satisfied or Furthermore, if the curvature value is greater than a preset threshold, it is determined to be a potential cylindrical transition surface; if And all of them are large values, so they are determined to be spherical transition surfaces; combined with the smoothness of the edges: if the surface contains at least two spring edges (i.e., the normal at the edge is continuous and connected to non-transition surfaces), then it is confirmed as a transition surface.

[0026] Furthermore, the minimum leaf subgraph priority strategy in step S3 has the following decision logic:

[0027] Define feature chain external connectivity Define the number of edges connecting the surface in the chain to surfaces outside the chain; define the feature chain. scale The total number of transition surfaces contained in the chain; if the feature chain satisfies and If a subgraph (with an adjustable threshold) does not contain complex T-shaped or Y-shaped self-intersecting structures, it is marked as a minimum leaf subgraph. The suppression order strictly follows: process all minimum leaf subgraphs first, and then process non-leaf subgraphs. When a minimum leaf subgraph is suppressed, if its parent feature chain meets the conditions, it is dynamically transformed into a new minimum leaf subgraph and added to the queue.

[0028] Furthermore, the topological suppression based on the Euler operator in step S4, specifically the processing flow for the edge-blend chain, is as follows:

[0029] Identify each transition surface in the chain. and its two sides of support surface Identify the two spring edges on the transition surface. And two intersecting edges or terminating edges; apply the KEF (Kill Edge Face) operator: delete and In the topological data structure and Merge into adjacent relationships and share the remaining edges; apply the KEV (Kill Edge Vertex) operator: if dangling or redundant vertices are generated after the transition surface is deleted, perform vertex collapse operation.

[0030] Furthermore, the implementation method of geometric stitching in step S4 is as follows:

[0031] After modifying the topology to support adjacent faces, obtain and Definition of underlying mathematical surfaces and The calculation is performed using either the Newton-Raphson method or the surface-to-surface intersection (SSI) algorithm. and Numerical intersection in Euclidean space ;Will As geometric data, it assigns the common edge of two supporting surfaces in the topology; if the intersection line If the intersection exceeds the boundary of the original model, the intersection line is trimmed based on the boundary loop (Wire) of the original model to ensure geometric consistency.

[0032] Furthermore, the incremental attribute adjacency graph local update mechanism in step S5 includes:

[0033] S51: Establish a list of affected nodes, initially containing the transition surface IDs that were deleted in step S3;

[0034] S52: Traverse the adjacent nodes of the nodes in the list of affected nodes and filter out all supporting surface nodes (i.e., the nodes corresponding to the supporting surfaces that are adjacent to the transition surface in step S3 and form a new adjacency relationship after topological modification).

[0035] S53: In the AAG data structure, physically delete the transition surface nodes and associated edges (including spring edges between transition surfaces and support surfaces, and cross edges between transition surfaces) in the list of affected nodes, and clear the invalid topology association information.

[0036] S54: Based on the geometric stitching results of step S4, check whether the geometric attributes of the support surface nodes selected in S52 have changed (e.g., the plane becomes a curved surface, or the area changes). If there are changes, update the attribute labels of the corresponding support surface nodes in the attribute adjacency graph synchronously.

[0037] S55: Based on the support surface topology adjacency relationship reconstructed in step S3 and the new common edge attributes (such as convex edges and concave edges) assigned in step S4, insert new edge connections between the support surface nodes selected in S52. The weight of the edge is set to the attribute of the new common edge, establishing an adjacency relationship consistent with the current model's topology and geometry. The time complexity of this process is O(n log n). ,in The number of nodes in a local region is much smaller than the number of nodes globally. .

[0038] Furthermore, the method also includes an exception handling rollback mechanism:

[0039] Before step S4 is executed, a snapshot of the current local topology is saved. If the geometric stitching calculation does not converge or produces self-intersections, the topology is rolled back using inverse Euler operators (MEF, MEV, etc.), and the feature chain is marked as requiring manual intervention. This feature is skipped and the next feature is processed to ensure that the automated process is not interrupted.

[0040] Furthermore, the transition surface includes edge transition surfaces and vertex transition surfaces: an edge transition surface is a surface that connects two non-transition surfaces; a vertex transition surface is a surface that connects three or more edge transition surfaces; adjacent transition surfaces are clustered into transition feature chains based on the cross edges in the attribute adjacency graph.

[0041] An apparatus for orderly suppression of transition features in complex models, comprising:

[0042] The model parsing module is used to parse the input 3D B-Rep model, extract point, edge, and face topological elements to construct an attribute adjacency graph, identify transition surfaces in the model by performing mesh sampling and differential geometry analysis on the surface, and map the transition surfaces to attribute labels of nodes in the attribute adjacency graph.

[0043] The transition feature clustering module is used to cluster adjacent transition surfaces into transition feature chains based on the edge attributes in the attribute adjacency graph.

[0044] The suppression queue planning module is used to analyze the external connectivity and nesting relationship of each transition feature chain in the attribute adjacency graph, identify the minimum leaf subgraph, and construct a priority suppression queue based on the minimum leaf subgraph; where the external connectivity is the number of edges in the transition feature chain that connect the surface to the surface of the non-chain.

[0045] The topology-geometric co-suppression module is used for topology suppression and geometric stitching based on Euler operators. It performs suppression on the current transition feature chain according to the priority suppression queue: it generates the corresponding Euler operator sequence according to the feature type, deletes the transition surface and modifies the topological connection relationship so that the support surfaces that were originally separated by the transition surface are topologically adjacent; then it calculates the intersection line of the underlying geometric surface of the support surface and updates the geometric information of the edges and vertices to close the model.

[0046] The incremental graph update module is used to lock the affected topological region after the suppression operation is completed, delete the failed node in the attribute adjacency graph and update the adjacent edges, and perform incremental local update of the attribute adjacency graph.

[0047] The iterative scheduling and control module is used to iterate steps S3 to S5 until all target transition feature chains are suppressed.

[0048] An electronic device includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement a method for orderly suppression of transition features of complex models.

[0049] In summary, the present invention has the following advantages:

[0050] Solving the problems of self-intersection and topological loopholes: Abandoning the traditional surface extension-clipping approach, the Euler operator is used to directly modify the topology and find the intersection and stitch the underlying surface, completely avoiding the self-intersection and divergence problems caused by free surface extension, and still maintaining the topological closure of the model at the intersection of Y-shaped / T-shaped features;

[0051] Significantly improves computational efficiency: The incremental local update mechanism reduces the graph update complexity from O(N²) to O(k), resulting in a significant improvement in efficiency for large-scale complex models;

[0052] Enhance robustness and process stability: The minimum leaf subgraph priority strategy avoids topological dependency conflicts, and the exception handling rollback mechanism can skip failure features, ensuring that the automated process is not interrupted and adapting to models with defects or complex convergence features.

[0053] Maintaining high-precision geometric boundaries: Based on the topology-geometry co-operation of the B-Rep model, the geometric accuracy is guaranteed by the precise surface intersection algorithm without losing the core shape features of the model, meeting the high-precision requirements of CAE finite element analysis and CAM machining trajectory planning;

[0054] High versatility: It does not rely on CAD design history, is suitable for intermediate format models such as STEP and IGES, and can handle complex models composed of various curved surfaces such as planes, cylinders, and NURBS, covering engineering application scenarios in industries such as aerospace and automotive. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating the suppression of transition features in the complex model of the present invention.

[0056] Figure 2(a) is a schematic diagram of the rounded corner transition feature chain T1 of the present invention;

[0057] Figure 2(b) is a schematic diagram of the adjacency graph of the T1 attribute of the present invention;

[0058] Figure 3(a) is a schematic diagram of transition feature suppression of the present invention including the rounded corner transition feature model T2;

[0059] Figure 3(b) is a schematic diagram of the transition feature suppression of the outer leaf subgraph preferentially processed by the present invention;

[0060] Figure 3(c) is a schematic diagram of the transition feature suppression of the inner leaf subgraph processed by the present invention;

[0061] Figure 4(a) is a schematic diagram of the transition surface suppression of the rounded corner transition feature S1 of the present invention;

[0062] Figure 4(b) is a schematic diagram of the transition surface suppression after the topology is modified by the S1 Euler operator of the present invention;

[0063] Figure 4(c) is a schematic diagram of transition surface suppression after S1 geometric normalization in this invention;

[0064] Figure 5(a) is a schematic diagram of the local update of the attribute adjacency graph of the present invention, which includes the rounded corner transition feature model T3;

[0065] Figure 5(b) is a schematic diagram of a partial update of the T3 attribute adjacency graph of the present invention;

[0066] Figure 5(c) is a schematic diagram of the local update of model T4 after suppressing some rounded corner transition features according to the present invention;

[0067] Figure 5(d) is a schematic diagram of a partial update of the T4 attribute adjacency graph of the present invention. Detailed Implementation

[0068] The present invention will now be described in further detail.

[0069] like Figure 1 The diagram illustrates an ordered suppression method for transition features in complex models based on the Euler operator and local graph updates, comprising:

[0070] Step S1: Constructing the attribute adjacency graph and feature chain recognition. In the B-Rep model, geometric and topological information are separated. This step aims to establish a mapping graph between geometry and topology. First, traverse all faces of the model. For each face... In the parameter domain Upper division The sampling grid is used to calculate the principal curvature of each sampling point. and Define the rules for determining face attributes:

[0071] Edge transition surface (EBF): If the minimum principal radius of curvature of the surface is... Within the set threshold range (e.g.) ), and the surface along If the rate of change of curvature in a direction is close to 0 (i.e., ruled or tubular surface characteristics), and its two sides are smoothly connected (G1 continuous), it is labeled as EBF.

[0072] Vertex Transition Surface (VBF): If a surface connects to 3 or more EBFs and has spherical or toroidal characteristics, it is marked as a VBF.

[0073] Constructing an attribute adjacency graph ,in Represents face nodes, Represents edge connection.

[0074] Spring Edge: The edge connecting the transition surface and the support surface, marked as a dashed line in the diagram, with a weight of 1.

[0075] Cross Edge: The edge that connects two adjacent transition surfaces. It is marked as a solid line in the diagram and has a weight of 0.

[0076] Strong adjacency edge: An edge that connects two non-transition surfaces, with a weight set to infinity.

[0077] Based on the Connected Components algorithm, all EBF and VBF nodes connected by "cross edges" are clustered to form several independent transition feature chains, denoted as... As an example, Figure 2(a) shows a schematic diagram of feature chain T1, illustrating a complex corner structure formed by the convergence of three rounded transition features with different radii. Three red tubular surfaces converge at the corner, forming a complex Y-shaped topology. This structure is difficult to process in one go using traditional methods because they share the same freeform surface vertex transition surface. Figure 2(b) shows a schematic diagram of the attribute adjacency graph of T1, which visualizes the AAG data structure corresponding to T1. The blue elliptical nodes in the graph represent faces in the model (e.g., the NURBS surface node centered at ID: 598, and the surrounding connected cylindrical surface nodes at IDs: 565, 595, and 603). The lines connecting the nodes represent the adjacency relationships between faces, and the values ​​on the lines (e.g., w: 4.408e-15) represent the attribute weights of the edges. Through this graphical mapping, the complex geometric connections are transformed into a graph theory structure that is easily processed by computers. The central node (ID: 598) connects the three branches, clearly expressing the topological structure of the feature.

[0078] Step S2: Suppression order planning based on the minimum leaf subgraph priority strategy. This is one of the core steps of the invention, aiming to solve complex topological dependency problems. For each feature chain... Calculate its graph theory properties:

[0079] External Degree ): Statistical Chain All nodes in the middle and those that do not belong The total number of "spring edges" between nodes. This typically means that the feature chain is a simple, open-ended rounded bar, or an isolated feature that is closed at both ends but topologically connected to only two principal planes. This means that the feature chain is complexly connected to multiple surfaces, possibly a Y-type joint or a multi-faceted intersection area.

[0080] Establish a priority queue Traverse all feature chains; if the condition is satisfied... If it is not marked as a "dependency", then it is defined as a "minimum leaf subgraph" and pushed into the queue. For nested features (such as a small rounded corner superimposed on a large rounded corner), since the small rounded corner (outer feature) is usually attached to the large rounded corner (inner feature), the small rounded corner's... Smaller, with large rounded corners It is often larger (because it connects the small rounded corner and the bottom). Therefore, this strategy naturally guarantees the order of "small to large, outside to inside". When the queue After a feature in the code is suppressed, the system checks its adjacent feature chains and updates them accordingly. Value. If the external degree of an internal feature chain drops to 2, its state is transformed into a "minimum leaf subgraph" and dynamically added to the queue. Figures 3(a), 3(b), and 3(c) illustrate the ordered suppression process of model T2, which contains nested rounded corner features. As shown in Figure 3(a), the original model T2 contains two distinct transition features: an inner circle with a larger rounded corner feature (connecting the top and side surfaces) and an outer circle with a smaller rounded corner feature (connecting the inner circle's rounded corner and the base), with a topological connection between them. As shown in Figure 3(b), the outer circle leaf subgraph is processed first. Based on the minimum leaf subgraph strategy, the algorithm first identifies the smaller red feature chain in the outer circle as a leaf node in the current graph structure. Therefore, the outer circle feature is suppressed first. The figure shows that the outer circle rounded corner has been removed, and the model is restored to a sharp-edge state, while the large rounded corner feature of the inner circle remains unchanged. As shown in Figure 3(c), after the outer circle is processed, the topological environment of the inner circle feature is updated, becoming a new minimum leaf subgraph. The algorithm then suppresses it, ultimately restoring the entire boss structure to a right-angled feature without chamfers.

[0081] Step S3: Traditional methods use Boolean operations, which are computationally intensive and unstable. This embodiment uses the Euler operator to directly modify the topological pointer, operating on a half-edge data structure. Assume the feature chain being processed is a simple edge rounding with a transition surface. Its two sides are support surfaces. and . It consists of a loop containing four sides: two spring sides shared with the support surface. , and two end edges (or no end edges in the case of a closed loop).

[0082] Execute the KEF (Kill Edge Face) operator: delete from the topology table. and edge .original The occupied topological space disappears. and There was a logical "gap" between them, but in the half-structure, it was corrected by modification. The co-edge pointer makes The left neighbor points to The right adjacent face points to At this point, from a topological logic perspective, and They are already adjacent and share an edge. .

[0083] Execute the KEV (Kill Edge Vertex) operator: If isolated vertices or overlapping edges are generated during the deletion process (e.g., when dealing with spherical chamfers), call the KEV operator to remove redundant tiny edges and vertices, ensuring a clean topology.

[0084] Executing the KFMV (Kill Face Make Vertex) operator: For fully enclosed vertex chamfers (such as rounding the corners of a cube), after deleting the face, the original loop shrinks to a single point. The KFMV operator is then used to degenerate the face back to a vertex, preserving Euler's formula. balance.

[0085] Step S4: Geometry Normalization and Stitching Steps. As shown in Figure 4(a), the original feature S1 displays a typical three-sided converging rounded corner structure, with the blue highlighted area representing the vertex transition surface to be suppressed. Figure 4(b) shows the intermediate state after modifying the topology using the Euler operator. It can be seen that the transition surface has been physically deleted, resulting in a geometric hole in the model at that location. However, at the data structure level, the surrounding surfaces have been marked as adjacent; only their geometric boundaries have not yet extended and converged.

[0086] extract The underlying geometric surface and The underlying geometric surface Note that this extracts an undone, mathematically defined infinite surface. Construct a system of intersection equations: The Newton-Raphson iterative method is used to solve this nonlinear equation system. Using the centerline of the original chamfered surface as the initial value, the iteration converges rapidly to obtain a high-precision three-dimensional intersection curve. The calculated curve Extract the valid segment (determined by the endpoint range of the original feature chain). Assign this curve segment to the shared edge retained in step S3. Update the 3D coordinates of the two vertices at the ends of the shared edge to the endpoints of the intersection line. Recalculate. and The parametric curve (P-Curve) on the new edge ensures consistency of the parameter domain. Figure 4(c) shows the final state after stitching. The algorithm calculates the theoretical intersection points and lines of the three surrounding planes and updates the topology with this geometric information. The original holes are filled with perfect right-angle vertices and sharp edges, and the model achieves accurate restoration from rounded corners to right angles.

[0087] Step S5: Incremental AAG Local Update In order to support batch ordered suppression, the AAG graph needs to reflect the model state in real time.

[0088] Figures 5(a) to 5(d) illustrate the local update effect of the attribute adjacency graph during the suppression process of model T3. As shown in Figure 5(a), this is a model with multiple rounded corner transition features, and its corresponding graph structure is shown in Figure 5(b), containing a large number of FilletSurface (transition surface) nodes (e.g., IDs: 9548, 9512) and the edges connecting them. These nodes represent the currently existing transition features. The algorithm records the set of surface IDs deleted in step S3. and the set of support face IDs whose adjacency relationships have been modified. In the adjacency list or adjacency matrix of AAG, remove directly The corresponding rows and columns have a time complexity of O(n). .examine The faces in the AAG. Based on the new topological relationship established in step S3, new solid edges (Rigid Edges) are inserted between these supporting face nodes. Only the continuity at the newly generated intersections (usually G0 or G1) needs to be checked, and the edge attributes are updated. As shown in Figure 5(c), after processing, the bottom rounded corners of the suppressed model T4 are removed, and the attribute adjacency graph shown in Figure 5(d) is obtained. Compared with Figure 5(b), it can be clearly seen that the graph structure has changed significantly. The original bottom FilletSurface nodes are removed, and in their place, direct connections are established between Plane nodes (such as ID: 589, 574) and Cylinder nodes. This update is local, and unaffected nodes (such as nodes far from the operation area) remain unchanged, fully demonstrating the efficiency of local updates. Compared with the traditional method, which re-traverses the entire model for each feature deletion (complexity) (where N is the number of model faces). This method only involves local updates. Each side ( This enables near real-time graph updates.

[0089] Although this method is highly stable, geometric stitching may fail in extreme cases (such as inherent design flaws or non-intersecting underlying surfaces). This embodiment introduces a transaction rollback mechanism: before executing step S3, a snapshot of the current local topology is saved. If, in step S4, the Newton iteration fails to converge within a preset number of iterations (e.g., 50), or the calculated intersection line is outside the model parameter domain, the system throws an exception. After catching the exception, the system calls the inverse Euler operator (MEF / MEV) to instantly restore the deleted faces and edges using the snapshot data, rolling the model back to its pre-operation state. Simultaneously, the feature chain is marked as "suppression failed - manual intervention required" in the AAG, and the next feature in the queue continues processing. This ensures that the automated process is not interrupted by individual bad features.

[0090] This invention, by combining the topological rigidity operation of the Euler operator with the intersection calculation of the underlying geometry, and with the graph theory-driven ordered processing strategy, fundamentally overcomes the geometric instability and topological conflict problems in the feature suppression of non-parametric freeform surface models, and provides a high-quality simplified model for CAE mesh generation and CAM machining trajectory planning.

[0091] The present invention also provides an apparatus for an ordered suppression method of transition features in complex models, comprising:

[0092] The model parsing module is used to parse the input 3D B-Rep model, extract point, edge, and face topological elements to construct an attribute adjacency graph, identify transition surfaces in the model by performing mesh sampling and differential geometry analysis on the surface, and map the transition surfaces to attribute labels of nodes in the attribute adjacency graph.

[0093] The transition feature clustering module is used to cluster adjacent transition surfaces into transition feature chains based on the edge attributes in the attribute adjacency graph.

[0094] The suppression queue planning module is used to analyze the external connectivity and nesting relationship of each transition feature chain in the attribute adjacency graph, identify the minimum leaf subgraph, and construct a priority suppression queue based on the minimum leaf subgraph; where the external connectivity is the number of edges in the transition feature chain that connect the surface to the surface of the non-chain.

[0095] The topology-geometric co-suppression module is used for topology suppression and geometric stitching based on Euler operators. It performs suppression on the current transition feature chain according to the priority suppression queue: it generates the corresponding Euler operator sequence according to the feature type, deletes the transition surface and modifies the topological connection relationship so that the support surfaces that were originally separated by the transition surface are topologically adjacent; then it calculates the intersection line of the underlying geometric surface of the support surface and updates the geometric information of the edges and vertices to close the model.

[0096] The incremental graph update module is used to lock the affected topological region after the suppression operation is completed, delete the failed node in the attribute adjacency graph and update the adjacent edges, and perform incremental local update of the attribute adjacency graph.

[0097] The iterative scheduling and control module is used for cyclic iteration until all target transition feature chains are suppressed.

[0098] An electronic device includes a processor and a memory, wherein a computer program is stored in the memory, and the processor executes the computer program to implement the above-mentioned method for orderly suppression of transition features of complex models.

[0099] This application forms a closed-loop collaborative system of data foundation, sequence guidance, topology modification, geometric matching, efficient update, and fault tolerance guarantee. Each feature is interdependent and mutually empowering, resulting in a significant synergistic effect. It successfully solves the technical bottlenecks in existing technologies, such as geometric instability, low efficiency, topology conflicts, and easy process interruption, which cannot be overcome by individual breakthroughs.

[0100] I. The core manifestation of synergy

[0101] 1. Foundation Layer Collaboration: The Attribute Adjacency Graph (AAG) provides a data foundation for all subsequent core operations.

[0102] Collaborative Logic: AAG deeply associates the topological elements (points, edges, faces) of the model with geometric properties (curvature, edge type: spring edge / cross edge), providing graph theory analysis basis such as external connectivity and scale for priority ranking of minimum leaf subgraphs, providing judgment criteria for the location of affected nodes for incremental local updates, and providing adjacency relationship data of transition surfaces and support surfaces for topological suppression of Euler operators.

[0103] Collaborative effect: As a unified data carrier, AAG eliminates the need for repeated model parsing in subsequent operations, enabling one-time construction and multiple reuses, laying the foundation for the efficiency of the overall process.

[0104] 2. Planning layer collaboration: Minimum leaf subgraph priority sorting provides an execution criterion for Euler operators.

[0105] Collaborative logic: The sorting result of the minimum leaf subgraph (outer first, inner second; smaller first, larger second) directly guides the execution order of the Euler operator, prioritizing feature chains with weak topological associations and small influence ranges to avoid subsequent operations from destroying the established topological relationships; at the same time, the sorting process relies on the graph theory properties (external connectivity, nesting relationship) provided by AAG, and the sorting result provides a dynamic update basis for iterative suppression (after the outer feature is suppressed, the graph theory properties of the inner feature are updated and added to the queue).

[0106] Synergistic effect: It solves the conflict problem caused by the disordered order of topology modification, and makes the topology operation of Euler operator change from blind execution to orderly and controllable, which greatly improves the robustness of complex nested model suppression.

[0107] 3. Core operation collaboration: Euler operator (topology modification) and intersection of underlying surface (geometric stitching) realize topology-geometric double closed loop.

[0108] Collaborative Logic: The Euler operator first deletes the transition surface and reconstructs the topological adjacency relationship of the support surface (solving the self-intersection problem caused by surface extension), and then calculates the precise intersection line of the support surface through the intersection of the underlying surface (Newton-Raphson iteration method / SSI), supplementing the geometric data for the blank adjacent edges after the topological modification; at the same time, the result of geometric stitching (new common edge attributes, support surface boundary update) provides the update basis for subsequent AAG local updates;

[0109] Synergistic effect: It completely avoids the risk of self-intersection in freeform surface extension, while ensuring the effectiveness of model topology and geometric closure. It solves the core contradiction of topology and geometry disconnect in existing technologies and achieves high-precision suppression of freeform surface models.

[0110] 4. Efficiency Collaboration: Incremental AAG local updates and preceding operations form a closed loop for efficiency improvement.

[0111] Collaborative logic: The list of affected nodes for local updates comes directly from the preceding operations, including the deleted transition surface IDs (initial nodes), the reconstructed support surface adjacency relationships (filtering support surface nodes), and the results of geometric stitching (updating support surface attributes); local updates only perform the deletion of invalid data → synchronization of geometric attributes → reconstruction of adjacent edges for these nodes, without the need for global reconstruction; at the same time, the updated AAG provides accurate topological-geometric data for the next round of iteration suppression;

[0112] Synergistic effect: Reduces update complexity, and works with preceding operations to suppress a feature and update locally once, which greatly improves the suppression efficiency of large-scale complex models and solves the pain point of low efficiency of existing graph-based methods.

[0113] 5. Fault tolerance and collaboration: The exception handling rollback mechanism provides a safety net for the entire process.

[0114] Cooperative logic: The triggering condition of the rollback mechanism (geometric stitching non-convergence / self-intersection) depends on the execution result of step S4. The rollback operation (inverse Euler operator MEF / MEV) is based on the local topology snapshot saved before step S3. After rollback, the failed features are marked and skipped to ensure that subsequent features continue to be executed according to the priority queue.

[0115] Synergistic effect: It transforms the entire closed-loop process from being prone to interruption to being robust and controllable. Even in extreme abnormal scenarios (such as model design defects), it can ensure that automatic suppression does not interrupt the process, further enhancing the engineering practicality of the method.

[0116] II. Core Technological Effects of Synergistic Effects

[0117] Simultaneously overcoming geometric instability and topological conflicts: the synergy of sequential programming + Euler operator + geometric stitching not only avoids the self-intersection of free surface extensions, but also resolves the topological dependency conflicts of nested features.

[0118] Simultaneously improving efficiency and accuracy: The synergy between local updates and AAG, topology-geometric operations, while ensuring geometric accuracy (intersection of underlying surfaces), improves efficiency from the global reconstruction level to the local update level, breaking through the industry bottleneck that high accuracy inevitably leads to low efficiency;

[0119] Simultaneously adaptable to complex models and cross-format scenarios: AAG construction does not rely on design history (adapting to intermediate formats such as STEP / IGES), and combined with sequential planning and topology-geometry collaboration, it can handle complex models containing freeform surfaces, nested features, and T-shaped / Y-shaped intersection features, covering scenarios that are difficult to handle with existing technologies;

[0120] Achieving a closed-loop automated process: From feature recognition, sequential planning, topology modification, geometric stitching, local updates to anomaly rollback, each link supports the others, forming a fully automated process that requires no human intervention, solving the pain point of existing technologies that require manual intervention to handle conflicts / anomalies.

[0121] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for orderly suppression of transition features in complex models, characterized in that: Includes the following steps, Step S1: Analyze the input 3D B-Rep model, extract point, edge, and face topological elements to construct an attribute adjacency graph, identify the transition surfaces in the model by performing mesh sampling and differential geometry analysis on the surface, and map the transition surfaces to attribute labels of the nodes in the attribute adjacency graph. Step S2: Based on the edge attributes in the attribute adjacency graph, cluster adjacent transition surfaces into transition feature chains; Step S3: Analyze the external connectivity and nesting relationship of each transition feature chain in the attribute adjacency graph, identify the minimum leaf subgraph, and construct a priority suppression queue based on the minimum leaf subgraph; where the external connectivity is the number of edges in the transition feature chain that connect the surface to the surface of the non-chain. Step S4: Based on the topological suppression and geometric stitching of the Euler operator, the current transition feature chain is suppressed according to the order of the priority suppression queue: the corresponding Euler operator sequence is generated according to the feature type, the transition surface is deleted and the topological connection relationship is modified so that the support surfaces that were originally separated by the transition surface are topologically adjacent; then the intersection line of the bottom geometric surface of the support surface is calculated, and the geometric information of the edges and vertices is updated to close the model. Step S5: After the suppression operation is completed, lock the affected topology region, delete the failed nodes and update the adjacent edges in the attribute adjacency graph, and perform incremental local update of the attribute adjacency graph. Step S6: Iterate through steps S3 to S5 until all target transition feature chains are suppressed.

2. The method for orderly suppression of transition features in complex models according to claim 1, characterized in that: The transition surface identification method in step S1 specifically includes: For any surface, in the parameter domain Internal partitioning The sampling grid is defined; the maximum principal curvature at each sampling point is calculated. and minimum principal curvature If satisfied or Furthermore, if the curvature value is greater than a preset threshold, it is determined to be a potential cylindrical transition surface; if And all of them are large values, so it is determined to be a spherical transition surface; combined with the smoothness of the edges: if the surface contains at least two spring edges, it is confirmed to be a transition surface.

3. The method for orderly suppression of transition features in complex models according to claim 1, characterized in that: The minimum leaf subgraph priority strategy in step S3 has the following decision logic: Define feature chain external connectivity Define the number of edges connecting the surface in the chain to surfaces outside the chain; define the feature chain. scale This represents the total number of transition surfaces contained in the chain; If the feature chain satisfies and If a subgraph does not contain complex T-shaped or Y-shaped self-intersecting structures, it is marked as a minimum leaf subgraph. The suppression order strictly follows: process all minimum leaf subgraphs first, and then process non-leaf subgraphs. When a minimum leaf subgraph is suppressed, if its connected parent feature chain meets the conditions, it is dynamically transformed into a new minimum leaf subgraph and added to the queue.

4. The method for orderly suppression of transition features in complex models according to claim 1, characterized in that: The topological suppression based on the Euler operator in step S4, specifically the processing flow for edge transition feature chains, is as follows: Identify each transition surface in the chain. and its two sides of support surface ; Identify the two spring edges on the transition surface And two intersecting edges or terminating edges; apply the KEF operator: delete and In the topological data structure and Merge into adjacent relationships and share the remaining edges; apply the KEV operator: if dangling or redundant vertices are generated after the transition surface is deleted, perform vertex collapse operation.

5. The method for orderly suppression of transition features in complex models according to claim 4, characterized in that: The implementation method of geometric stitching in step S4 is as follows: After modifying the topology to support adjacent faces, obtain and Definition of underlying mathematical surfaces and The Newton-Raphson iteration method or the surface-to-surface intersection algorithm is used to calculate... and Numerical intersection in Euclidean space ;Will As geometric data, it assigns the common edge of two supporting surfaces in the topology; if the intersection line If the intersection line exceeds the boundary of the original model, it is trimmed based on the boundary loop of the original model to ensure geometric consistency.

6. The method for orderly suppression of transition features in complex models according to claim 1, characterized in that: The incremental attribute adjacency graph local update mechanism in step S5 includes: S51: Establish a list of affected nodes, initially containing the transition surface IDs that were deleted in step S3; S52: Traverse the adjacent nodes of the nodes in the list of affected nodes and filter out all supporting face nodes; S53: In the attribute adjacency graph data structure, physically delete the transition surface nodes and associated edges in the list of affected nodes, and clear the invalid topological association information; S54: Based on the geometric stitching results of step S4, check whether the geometric attributes of the supporting surface nodes selected in S52 have changed. If there are changes, update the attribute labels of the corresponding supporting surface nodes in the attribute adjacency graph synchronously. S55: Based on the support surface topology adjacency relationships reconstructed in step S3 and the new common edge attributes assigned in step S4, insert new edge connections between the support surface nodes selected in S52, setting the edge weights to the attributes of the new common edges, and establishing adjacency relationships consistent with the current model's topology and geometry; the time complexity of this process is O(n log n). ,in The number of nodes in a local region is much smaller than the number of nodes globally. .

7. The method for orderly suppression of transition features in complex models according to claim 1, characterized in that: The method also includes an exception handling rollback mechanism: Before step S4 is executed, a snapshot of the current local topology is saved. If the geometric stitching calculation does not converge or produces self-intersections, the topology is rolled back using the inverse Euler operator, and the feature chain is marked as requiring manual intervention. The feature is skipped and the next feature is processed to ensure that the automated process is not interrupted.

8. The method for orderly suppression of transition features in complex models according to claim 1, characterized in that: Transition surfaces include edge transition surfaces and vertex transition surfaces: an edge transition surface is a surface that connects two non-transition surfaces; a vertex transition surface is a surface that connects three or more edge transition surfaces; adjacent transition surfaces are clustered into transition feature chains based on the cross edges in the attribute adjacency graph.

9. The apparatus for the method of ordered suppression of transition features in complex models according to any one of claims 1-8, characterized in that, include: The model parsing module is used to parse the input 3D B-Rep model, extract point, edge, and face topological elements to construct an attribute adjacency graph, identify transition surfaces in the model by performing mesh sampling and differential geometry analysis on the surface, and map the transition surfaces to attribute labels of nodes in the attribute adjacency graph. The transition feature clustering module is used to cluster adjacent transition surfaces into transition feature chains based on the edge attributes in the attribute adjacency graph. The suppression queue planning module is used to analyze the external connectivity and nesting relationship of each transition feature chain in the attribute adjacency graph, identify the minimum leaf subgraph, and construct a priority suppression queue based on the minimum leaf subgraph; where the external connectivity is the number of edges in the transition feature chain that connect the surface to the surface of the non-chain. The topology-geometric co-suppression module is used for topology suppression and geometric stitching based on Euler operators. It performs suppression on the current transition feature chain according to the priority suppression queue: it generates the corresponding Euler operator sequence according to the feature type, deletes the transition surface and modifies the topological connection relationship so that the support surfaces that were originally separated by the transition surface are topologically adjacent; then it calculates the intersection line of the underlying geometric surface of the support surface and updates the geometric information of the edges and vertices to close the model. The incremental graph update module is used to lock the affected topological region after the suppression operation is completed, delete the failed node in the attribute adjacency graph and update the adjacent edges, and perform incremental local update of the attribute adjacency graph. The iterative scheduling and control module is used to perform loop iterations until all target transition feature chains are suppressed.

10. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the method for orderly suppression of transition features of complex models as described in any one of claims 1-8.

Images