A groove type machining feature recognition method and system

Abstract

The present application relates to a kind of slot processing feature identification method and system.The present application includes input with slot feature parts three-dimensional geometric model;Extract the edge adjacency relationship of each face in the three-dimensional geometric model, construct adjacency graph, generate the attribute adjacency matrix of model based on the adjacency graph;Slot database is constructed;Based on the slot database, the face edge adjacency relationship of each slot feature template is extracted, and the corresponding pre-matching subgraph matrix is constructed;Based on the pre-matching subgraph matrix, from the attribute adjacency matrix of the model, the candidate slot feature that is consistent with each feature subgraph topological structure is screened out;The candidate slot feature is sequentially screened and classified according to the preset order;According to the preset face and edge geometric rule, the geometric constraint verification is carried out to candidate slot feature;Based on the final slot feature set, the slot processing feature in three-dimensional geometric model is output.The present application effectively solves the technical problem that the effect of conventional method is poor in common slot feature identification.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and in particular to a method and system for identifying the processing features of grooves. Background Technology

[0002] With the widespread application of Computer-Aided Design (CAD) and Computer-Aided Manufacturing (CAM) technologies, the integration of product development and manufacturing is constantly improving. As a key link connecting CAD and CAM, Computer-Aided Process Planning (CAPP) needs to automatically extract machining features with engineering semantics from 3D models, such as holes, slots, and cavities, to support process planning and automatic CNC program generation. Among these, machining feature recognition is one of the fundamental technologies for realizing the integration of CAD-CAPP-CAM and intelligent CNC programming.

[0003] Currently, for features with relatively simple geometric shapes such as holes and chamfers, there are mature recognition methods and commercial software implementations that can be stably applied in engineering practice. However, groove features are usually composed of multiple surfaces and have diverse shapes (such as closed grooves, open grooves, wide-body grooves, and arc-side grooves). Furthermore, complex topological phenomena such as cylindrical fragments and smooth edges are easily generated during the conversion of 3D model formats. This leads to existing recognition methods based on simple rules or local geometric criteria being prone to omissions and misidentifications on complex parts, making it difficult to achieve unified, accurate, and robust recognition of groove features. This problem directly restricts the automation level of groove machining process planning and also affects the intelligence level of subsequent toolpath planning and machining simulation.

[0004] In existing technologies, some methods identify machining features based on attribute adjacency graphs or deep learning, but these methods mostly remain at the level of general feature classification, failing to accurately extract the complete geometric attributes of groove features and lacking the ability to handle practical engineering problems such as model topological fragmentation. For example, patent CN110795797B (MBD Model Analysis) achieves graph structure matching of features through attribute adjacency graphs, but often results in mismatches when processing non-standardized features and cannot handle the fragmented surface problem that occurs during the 3D model conversion process. Another example is the method proposed by Li Wang et al. (2023) of Hunan University of Science and Technology in "A Method for Recognizing Machining Features of Irregular Parts Based on Graph and Volume Decomposition." Although it can handle some intersecting features, it still suffers from insufficient recognition accuracy when processing complex groove features, and it relies on Boolean operations for volume decomposition, resulting in high computational complexity and poor real-time performance. Therefore, existing technologies mainly have the following shortcomings:

[0005] (1) It is difficult to identify and process the geometric composition and diversity of various groove features in complex three-dimensional models, especially the accurate identification of complex groove types such as closed grooves, wide grooves, and arc-edge closed grooves;

[0006] (2) There is a lack of effective processing mechanism for the broken surface problem generated during the conversion of 3D CAD models, which makes it impossible to reconstruct surface clusters, resulting in a decrease in the accuracy of groove feature recognition;

[0007] (3) It relies heavily on predefined rules or large-scale training data, lacks the ability to analyze complete topological features, and is difficult to adapt to complex and ever-changing engineering needs.

[0008] (4) It is impossible to output a complete set of groove feature geometric parameters that meet manufacturing requirements, which limits the level of automation of subsequent process design and processing path planning. Summary of the Invention

[0009] To address this issue, the present invention provides a method and system for identifying groove processing features. This system, starting from the fundamental constituent surfaces and geometric constraints of groove features, possesses the ability to uniformly identify multiple types of groove features, supports fragmented surface preprocessing and surface cluster reconstruction, and can extract complete geometric parameters—an automated identification system. This effectively solves the technical problem of traditional methods performing poorly in identifying common groove features.

[0010] To solve the above technical problems, the present invention provides a method for identifying groove processing features, comprising:

[0011] Input a 3D geometric model of a part with groove-like features;

[0012] The edge adjacency relationships of each face in the three-dimensional geometric model are extracted, an adjacency graph is constructed, and an attribute adjacency matrix of the model is generated based on the adjacency graph. The fragmented faces generated by the format conversion are identified and merged in the preprocessing. Faces with common axes and consistent geometric parameters are reconstructed into complete face clusters to ensure the continuity of the topological structure.

[0013] Construct a slot class database, which includes slot class feature templates for: simple wide-body slots, bottom-closed slots, bottomless closed slots, bottom-open slots, and bottomless open slots;

[0014] Based on the slot database, the face-edge adjacency relationships of each slot feature template are extracted to construct the corresponding pre-matching subgraph matrix; wherein, the face-edge adjacency relationships are represented by matrix representation of face type encoding and edge adjacency encoding;

[0015] Based on the pre-matched subgraph matrix, candidate slot features that are consistent with the topology of each feature subgraph are selected from the attribute adjacency matrix of the model.

[0016] The candidate slot features are systematically filtered and categorized according to a preset order;

[0017] Based on the preset geometric rules of faces and edges, the candidate slot features are geometrically constrained and verified. Redundant instances that do not meet the geometric constraints and overlap in spatial position are eliminated to obtain the final slot feature set.

[0018] Based on the final groove feature set, the groove machining features in the three-dimensional geometric model are output.

[0019] In one embodiment of the present invention, the edge adjacency relationships of each face in the three-dimensional geometric model are extracted to construct an adjacency graph, including:

[0020] Each face of the three-dimensional geometric model is labeled with a face ID, and adjacency graph nodes are created using the face IDs.

[0021] For any shared edge in the three-dimensional geometric model, establish a bidirectional topological association connecting the two sides of the edge to form an adjacency graph containing directed edges and adjacency graph nodes.

[0022] In one embodiment of the present invention, generating an attribute adjacency matrix of the model based on the adjacency graph includes:

[0023] Encode the face attributes into the diagonal elements of the attribute adjacency matrix, wherein the face attributes include face geometry type, normal vector, and area;

[0024] Edge attributes are encoded into the off-diagonal elements of the attribute adjacency matrix, and the edge attributes include length, concavity and smoothness, inner and outer loop markings, and edge geometry type.

[0025] In one embodiment of the present invention, the diagonal elements are encoded using surface type, and the five primitives are defined as follows: 1 for plane, 2 for cylindrical surface, 3 for conical surface, 4 for spherical surface, and 5 for torus.

[0026] The off-diagonal elements are encoded using adjacency codes, and the encoding rules are as follows:

[0027] code = X×100 + Y×10 + Z;

[0028] Where X∈{1,2}, 1 and 2 represent the outer ring and the inner ring respectively;

[0029] Y∈{1,2,3}, where 1, 2, and 3 represent concave edge, convex edge, and smooth edge, respectively.

[0030] Z∈{1,2,3,4}, where 1, 2, 3, and 4 represent the sides as a straight line, a circle, a closed curve, and an ellipse, respectively.

[0031] In one embodiment of the present invention, the fragmented surfaces generated by format conversion are identified and merged in a preprocessing step, and the facets with a common axis and consistent geometric parameters are reconstructed into a complete face cluster to ensure the continuity of the topological structure, including:

[0032] After constructing the adjacency graph, while maintaining the face adjacency relationship, the nodes that are coaxial and have the same diameter, belong to the same cylindrical analytical surface but are split into multiple fragments are clustered into the same face cluster, based on the consistency of the cylindrical geometric parameters.

[0033] Subsequently, the graph is folded and reconstructed using face clusters as the basic unit. Representative face nodes are selected within the face cluster, and the external adjacent edges of each face within the face cluster are merged. Furthermore, the external edges pointing to any member within the cluster are uniformly redirected to the representative node, thereby constructing a face cluster adjacency graph with node semantics of "complete cylindrical face" and edge semantics inheriting the original adjacency attributes.

[0034] In one embodiment of the present invention, based on the slot class database, the face-edge adjacency relationships of each slot class feature template are extracted to construct a corresponding pre-matched subgraph matrix, including:

[0035] Starting from the structural definition of the feature templates of each type of groove, each type of groove is decomposed into several component surfaces with clear processing semantics. The component surfaces include the bottom surface, two side surfaces, end arc surface or end conical arc surface.

[0036] The face-edge topological adjacency relationships between the constituent faces are extracted as the edge set of the template subgraph. The nodes are encoded using face type, which includes plane, cylindrical, and conical surfaces. The edges are encoded using edge adjacency encoding to encode the contact or transition attributes between adjacent faces. The contact or transition attributes include concave edges, convex edges, smooth edges, and direction combination information.

[0037] By treating non-adjacent relationships as explicit missing edge constraints, the face-edge adjacency relationships of the slot-type features are normalized into a computable and comparable graph structure description.

[0038] Based on the graph structure description, each slot template is encoded as the pre-matching subgraph matrix, wherein the diagonal elements of the pre-matching subgraph matrix are the face type codes of each constituent face, and the off-diagonal elements of the pre-matching subgraph matrix are the face edge adjacency codes. When two constituent faces are not adjacent, the corresponding element is -1.

[0039] To address the edge adjacency encoding variants that occur in actual modeling, including inner loop label compatibility, adjacency direction reversal, and smooth edge difference, multiple compatibility template pre-matching subgraph matrices are constructed and written into the slot class database to form a feature database for subsequent graph matching.

[0040] In one embodiment of the present invention, based on the pre-matched subgraph matrix, candidate slot features consistent with the topological structure of each feature subgraph are selected from the attribute adjacency matrix of the model, including:

[0041] The pre-matched subgraph matrix corresponding to each slot feature template is taken as matrix A, and the adjacency graph extracted from the three-dimensional geometric model is taken as graph B.

[0042] The matrix A is converted into an attribute adjacency graph structure, where the diagonal elements represent the face types and the off-diagonal elements represent the adjacency relationships between faces.

[0043] A depth-first search algorithm is used to search for candidate mapping results in graph B that satisfy the subgraph isomorphic mapping conditions, starting from the non-fragmented nodes in matrix A. The subgraph isomorphic mapping conditions include: exact matching of node values, completely consistent edge weights, and maintaining the explicitly specified adjacency and non-adjacency constraints in matrix A.

[0044] The impact of fragmented surface merging is addressed during the matching process. By mapping the representative surface to the original fragmented surface, multiple fragmented surfaces in the B diagram can be mapped to a single representative surface in the A matrix, ensuring that fragmented surfaces do not affect the topology consistency judgment.

[0045] After matching is completed, the candidate mapping results are normalized and cleaned, invalid candidates containing duplicate face indices or face set conflicts are removed, and a unique set of candidate slot feature instances is output as the candidate slot features.

[0046] In one embodiment of the present invention, the candidate slot features are systematically screened and categorized according to a preset order, including:

[0047] Prioritize identifying closed groove feature templates and arc-side closed groove feature templates, then identify bottomless closed groove feature templates and arc-side bottomless closed groove feature templates, then identify open groove feature templates, and finally identify simple wide-body groove feature templates.

[0048] In one embodiment of the present invention, geometric constraint verification is performed on the candidate slot features according to preset geometric rules for faces and edges, including:

[0049] For each type of slot feature template, obtain the candidate mapping results that satisfy the template topological constraints, and restore the set of faces involved in the candidate instances to the original set of faces based on the mapping relationship between the representative face and the original fragment face;

[0050] The geometric rules corresponding to the slot type are invoked for geometric constraint verification. Among them, the geometric rules for closed slots include verification of the relative positional relationship between the bottom and the side, verification of the parallelism of the side, verification of the coaxiality of the cylindrical side, and verification of the radius consistency. The geometric rules for bottomless closed slots include performing cavity semantic judgment to exclude false features of the boss type when the bottom is missing. The geometric rules for open slots include detecting the extension intersection of the side along the opening direction to exclude non-slot structures that intersect with the solid.

[0051] After the geometric constraint verification is passed, deduplication is performed based on the set of occupied faces. If a new candidate instance has an intersection with the face set of a retained instance, it is removed; otherwise, it is included in the final result and the set of occupied faces is updated.

[0052] In the two-round recognition process of prioritizing merging the graph and supplementing the original graph, the consistent update logic of the occupied face set is maintained, so that the deduplication and retention decisions follow the preset order, thereby obtaining a final set of slot class feature instances that do not overlap and have consistent priority, which serves as the final slot class feature set.

[0053] The present invention also provides a groove processing feature recognition system, comprising:

[0054] The 3D geometric model input module is used to input the 3D geometric model of a part with groove-like features;

[0055] The attribute adjacency matrix generation module is used to extract the edge adjacency relationships of each face in the three-dimensional geometric model, construct an adjacency graph, generate the attribute adjacency matrix of the model based on the adjacency graph, and identify and merge the fragmented faces caused by format conversion. The face patches with common axes and consistent geometric parameters are reconstructed into complete face clusters to ensure the continuity of the topological structure.

[0056] The slot database construction module is used to construct a slot database, which includes slot feature templates for: simple wide-body slots, bottom-closed slots, bottomless closed slots, bottom-open slots, and bottomless open slots.

[0057] The pre-matching subgraph matrix construction module is used to extract the face-edge adjacency relationship of each slot class feature template based on the slot class database and construct the corresponding pre-matching subgraph matrix; wherein, the face-edge adjacency relationship is represented by a matrix of face type encoding and edge adjacency encoding;

[0058] The candidate slot feature filtering module is used to filter out candidate slot features that are consistent with the topological structure of each feature subgraph from the attribute adjacency matrix of the model based on the pre-matched subgraph matrix.

[0059] An ordered filtering and classification module is used to perform ordered filtering and classification of the candidate slot features according to a preset order;

[0060] The final slot feature set acquisition module is used to perform geometric constraint verification on the candidate slot features according to the preset geometric rules of the face and edge, and remove redundant instances that do not meet the geometric constraints and have overlapping spatial positions to obtain the final slot feature set.

[0061] The groove machining feature output module is used to output the groove machining features in the three-dimensional geometric model based on the final groove type feature set.

[0062] The technical solution of the present invention has the following advantages compared with the prior art:

[0063] The present invention discloses a method and system for identifying groove machining features. Based on a database of common groove types and an attribute adjacency matrix, and combined with steps such as fragment merging preprocessing, feature subgraph matching, and geometric constraint verification, it can stably and accurately identify a variety of typical groove machining features on a 3D model with fragments and complex topological relationships. This effectively improves the automation and robustness of feature recognition and solves the technical problems of high false negative rate and poor generalization ability in the identification of groove features of complex parts by traditional methods.

[0064] This invention constructs an adjacency graph by extracting the edge adjacency relationships of each face in a 3D geometric model, and further generates an attribute adjacency matrix. Diagonal elements encode face attributes (face geometry type, normal vector, area), while off-diagonal elements encode edge attributes (length, concavity / convexity, smoothness, inner / outer loop markings, edge geometry type), avoiding the strong reliance on local geometric thresholds and manual rules found in traditional methods. Through the matrix representation of face type encoding and edge adjacency encoding, it can constrain which faces constitute the slot and how faces are adjacent at the topological level, and distinguish concave edges, convex edges, smooth edges, and boundary loop attributes at the semantic level, improving the stability and consistency of slot feature recognition.

[0065] This invention addresses the complex topological phenomena such as fragmented surfaces and smooth edges on cylinders caused by format conversion from CAD to STEP. The invention proposes a fragmented surface identification and merging preprocessing: using the consistency of cylindrical geometric parameters as a criterion, multiple fragmented surface nodes that are coaxial, have the same diameter, belong to the same cylindrical analytical surface but have been split are clustered into the same surface cluster, and the adjacency graph is reconstructed by folding the surface cluster as the basic unit to ensure the continuity of the topological structure.

[0066] This invention uses the template pre-matched subgraph matrix as matrix A and the model adjacency graph as graph B. A depth-first search is employed in graph B to search for candidate mapping results that satisfy the subgraph isomorphism mapping conditions. Matching conditions include exact node value matching, completely consistent edge weights, and preservation of adjacency and non-adjacency constraints in the template. This significantly reduces the probability of false features entering the candidate set based solely on local concave edges or local parallelism. During the matching process, a fragmented surface merging process is introduced to handle the impact, allowing multiple fragmented surfaces to be mapped to a single representative surface in the template. This ensures that fragmented surfaces do not disrupt the topological consistency judgment, thereby maintaining recognition stability in the engineering model.

[0067] This invention systematically filters and classifies candidate slot features according to a preset order, employing an identification strategy of "closed slots first, then open slots; strict identification first, then lenient identification; and wide-body slots last." It prioritizes identifying closed slots and arc-sided closed slots, followed by bottomless closed slots, then open slots, and finally simple wide-body slots. This reduces duplicate identification and misclassification.

[0068] The present invention further verifies the candidate slot features by geometric constraints based on preset geometric rules of surfaces and edges, which can effectively reduce the false recognition rate. Attached Figure Description

[0069] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0070] Figure 1 This is a flowchart of the groove processing feature recognition method of the present invention.

[0071] Figure 2 It is a three-dimensional model diagram containing a closed groove.

[0072] Figure 3 This is a schematic diagram of a partial adjacency matrix of the model.

[0073] Figure 4 This is a schematic diagram of the merging of fragmented surfaces.

[0074] Figure 5 This is a schematic diagram of a simple wide-body slot and its corresponding pre-matching subgraph matrix.

[0075] Figure 6 It is a schematic diagram of a bottom closed groove and its corresponding pre-matching subgraph matrix.

[0076] Figure 7 This is a schematic diagram of a bottomless closed groove and its corresponding pre-matching subgraph matrix.

[0077] Figure 8 This is a schematic diagram of a bottomed open slot and its corresponding pre-matched subgraph matrix.

[0078] Figure 9 This is a schematic diagram of a bottomless open slot and its corresponding pre-matched subgraph matrix.

[0079] Figure 10 This is a flowchart of the matching process for common slot features.

[0080] Figure 11 This is a diagram of the final recognition result. Figure 1 .

[0081] Figure 12 It is a schematic diagram of a model containing various common groove processing features.

[0082] Figure 13 It is the constructed model diagram.

[0083] Figure 14 It is the adjacency matrix graph of the model (selected from the ten nodes of the model).

[0084] Figure 15 This is a diagram of the final recognition result. Figure 2 .

[0085] Figure 16 The display results are for the characteristics of each type of slot. Figure 1 .

[0086] Figure 17 The display results are for the characteristics of each type of slot. Figure 2 .

[0087] Figure 18 The display results are for the characteristics of each type of slot. Figure 3 .

[0088] Figure 19 The display results are for the characteristics of each type of slot. Figure 4 .

[0089] Figure 20 The display results are for the characteristics of each type of slot. Figure 5 .

[0090] Figure 21 The display results are for the characteristics of each type of slot. Figure 6 .

[0091] Figure 22 The display results are for the characteristics of each type of slot. Figure 7 . Detailed Implementation

[0092] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0093] In this invention, when directions (up, down, left, right, front, and back) are described, it is only for the convenience of describing the technical solution of this invention, and does not indicate or imply that the technical features referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0094] In this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," "exceeding," etc., are understood to exclude the stated number; "above," "below," "within," etc., are understood to include the stated number. In the description of this invention, the terms "first" and "second" are used only to distinguish technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0095] In this invention, unless otherwise explicitly defined, the terms "setting," "installing," and "connecting" should be interpreted broadly. For example, they can refer to a direct connection or an indirect connection through an intermediate medium; a fixed connection, a detachable connection, or an integrally formed connection; a mechanical connection, an electrical connection, or a connection capable of mutual communication; or the internal connection of two components or the interaction between two components. Those skilled in the art can reasonably determine the specific meaning of the above terms in this invention based on the specific content of the technical solution.

[0096] Example 1

[0097] Reference Figure 1 As shown, this embodiment provides a method for identifying groove processing features, including:

[0098] S1. Input a three-dimensional geometric model of a part with groove-like features.

[0099] S2. Extract the edge adjacency relationships of each face in the three-dimensional geometric model, construct an adjacency graph, generate the attribute adjacency matrix of the model based on the adjacency graph, and preprocess the fragmented faces generated by the format conversion by identifying and merging them, and reconstruct the face patches with common axes and consistent geometric parameters into complete face clusters to ensure the continuity of the topological structure.

[0100] Specifically, the edge adjacency relationships of each face in the three-dimensional geometric model are extracted to construct an adjacency graph, including:

[0101] S21. Label each face of the three-dimensional geometric model as a face ID, and create adjacency graph nodes using the face IDs;

[0102] S22. For any shared edge in the three-dimensional geometric model, establish a bidirectional topological association connecting the two sides of the edge to form an adjacency graph containing directed edges and adjacency graph nodes.

[0103] For example, each face of the 3D model is labeled with a face ID [0,1,2,…,n], and n adjacency graph nodes are created accordingly (the node ID is the same as the face ID). All common edges are retrieved from the model, totaling K. Since each edge connects exactly two faces, a bidirectional topological association is established for each physical edge in the adjacency graph, forming 2K directed edges: if edge L connects face i and face j, it is denoted as the directed edge {Lij} from node i to node j and the directed edge {Lji} from node j to node i (i≠j). Thus, the adjacency graph corresponding to this model has n nodes and 2K directed edges.

[0104] Specifically, based on the adjacency graph, the attribute adjacency matrix of the model is generated, including:

[0105] S23. In addition to this pure topology, in order to ensure the accuracy and robustness of recognition, the adjacency graph must simultaneously carry surface attributes and edge attributes: the surface attributes are encoded into the diagonal elements of the attribute adjacency matrix, and the surface attributes include the surface geometry type (plane, cylinder, cone, sphere, torus, etc.), normal vector and area.

[0106] S24. Encode the edge attributes into the off-diagonal elements of the attribute adjacency matrix, wherein the edge attributes include length, concavity and smoothness, inner and outer ring markings (the inner and outer relationships of two faces on a shared boundary ring), and edge geometry type (straight line, circle, closed curve, ellipse).

[0107] Specifically, the off-diagonal elements are encoded using adjacency codes. The three semantic categories—face geometry type, normal vector, and area—are compressed into an integer adjacency code and written into the off-diagonal elements of the matrix. The encoding rule is as follows:

[0108] code = X×100 + Y×10 + Z;

[0109] Where X∈{1,2}, 1 and 2 represent the outer ring and the inner ring respectively;

[0110] Y∈{1,2,3}, where 1, 2, and 3 represent concave edge, convex edge, and smooth edge, respectively.

[0111] Z∈{1,2,3,4}, where 1, 2, 3, and 4 represent the sides as a straight line, a circle, a closed curve, and an ellipse, respectively.

[0112] For example, 131 represents the adjacency of "outer ring-smooth-straight line". In the specific implementation, the model is first traversed through a dedicated extractor to produce graph data (number of nodes, face attributes, edge list and edge attributes).

[0113] The diagonal elements are encoded using face type, following a convention of five primitive types: 1 for plane, 2 for cylinder, 3 for cone, 4 for sphere, and 5 for torus. Cross-validation and correction are performed against the actual type of the geometric kernel to improve consistency. For off-diagonal lines, the adjacency code is calculated for each edge according to the above encoding rules and written into the matrix, thus obtaining an n×n face attribute adjacency matrix.

[0114] Meanwhile, an adjacency graph structure is used for subsequent template matching and recognition: each face is a vertex, the vertex stores the face type information, and the weight of the directed edge stores the adjacency code.

[0115] Reference Figure 2 As shown, a 3D model containing closed slots exhibits multiple common slots of the same type. For ease of explanation, only a subset of these slots is analyzed. After extracting the face edge information of the model, a partial adjacency matrix is ​​constructed. Figure 3 As shown.

[0116] Specifically, the fragmented surfaces generated by format conversion are identified and preprocessed by merging, and the surfaces with common axes and consistent geometric parameters are reconstructed into complete surface families to ensure the continuity of the topology, including:

[0117] To address the common problem of cylindrical surface splitting in format conversion of 3D models (such as CAD to STEP / STP), this embodiment, after constructing the attribute face adjacency graph (AAG / AAM), uses the consistency of cylindrical geometric parameters as a criterion to cluster coaxial and equal-diameter nodes belonging to the same cylindrical analytical surface but split into multiple fragments into the same face cluster, while maintaining the face adjacency relationship.

[0118] Subsequently, the graph is folded and reconstructed using face clusters as the basic unit. Representative face nodes are selected within the face cluster, and the external adjacent edges of each face within the face cluster are merged. Furthermore, the external edges pointing to any member within the cluster are uniformly redirected to the representative node, thereby constructing a face cluster adjacency graph with node semantics of "complete cylindrical face" and edge semantics inheriting the original adjacency attributes.

[0119] Reference Figure 4 As shown, the above reconstruction effectively eliminates the graph node expansion and topological redundancy caused by cylinder splitting, enabling the adjacency relationship to be uniformly expressed at a more fundamental geometric level. At the same time, it establishes and preserves the index mapping from face cluster nodes to the original fragment set, laying the foundation for subsequent accurate geometric verification and result backfilling.

[0120] Furthermore, to ensure structural consistency between the model graph and the feature template subgraph during feature recognition, this embodiment also follows the aforementioned face clustering principle in defining or processing slot-type feature templates. This unifies the topological structure of the feature template from the "original face node mode" to a "face cluster node mode" that adapts to the reconstructed graph, thereby avoiding recognition failures caused by the mismatch between the model fragments and the ideal template topology. When performing feature subgraph matching based on the face cluster adjacency graph, the search space is significantly compressed, and the matching efficiency is greatly improved. After identifying candidate feature subgraphs, the representative nodes are expanded back to the original face set using the pre-stored face cluster-fragment mapping relationship to support subsequent geometric parameter extraction (such as fitting aperture and slot width) and engineering applications. This achieves an organic combination of "efficient and robust recognition based on face cluster graphs" and "accurate result output oriented towards the original face".

[0121] S3. Construct a slot type database, which includes slot type feature templates for: simple wide-body slots, bottom-closed slots, bottomless closed slots, bottom-open slots, and bottomless open slots.

[0122] S4. Based on the slot class database, extract the face-edge adjacency relationship of each slot class feature template and construct the corresponding pre-matching subgraph matrix; wherein, the face-edge adjacency relationship is represented by matrix representation of face type encoding and edge adjacency encoding.

[0123] Specifically, step S4 includes:

[0124] S41. Starting from the structural definition of each type of groove feature template, each type of groove is decomposed into several component surfaces with clear processing semantics. The component surfaces include bottom surface, two side surfaces, end arc surface or end conical arc surface.

[0125] S42. Extract the face-edge topological adjacency relationship between the constituent faces as the edge set of the template subgraph. The nodes are encoded using face type, which includes plane, cylindrical, and conical surfaces. The edges are encoded using edge adjacency to encode the contact or transition attributes between adjacent faces. The contact or transition attributes include concave edges, convex edges, smooth edges, and direction combination information.

[0126] S43. Treat the non-adjacent relationship as an explicit missing edge constraint, thereby normalizing the face-edge adjacency relationship of the slot-type feature into a computable and comparable graph structure description.

[0127] S44. Based on the graph structure description, each slot template is encoded into the pre-matching subgraph matrix, wherein the diagonal elements of the pre-matching subgraph matrix are the face type codes of each constituent face, and the off-diagonal elements of the pre-matching subgraph matrix are the face edge adjacency codes. When two constituent faces are not adjacent, the corresponding element is -1.

[0128] S45. For edge adjacency encoding variants that appear in actual modeling, the edge adjacency encoding variants include inner loop label compatibility, adjacency direction reversal, and smooth edge difference. Multiple compatibility template pre-matching subgraph matrices are constructed and written into the slot class database to form a feature database for subsequent graph matching.

[0129] It should be noted that this embodiment targets common groove processing features and establishes a unified representation method of "feature database - pre-matched subgraph matrix": First, starting from the structural definition of groove features, each type of groove is decomposed into several component surfaces with clear processing semantics (such as bottom surface, two side surfaces, end arc surface / conical arc surface, etc.), and the surface-edge topological adjacency relationship between these component surfaces is extracted as the edge set of the template subgraph; among them, nodes are encoded by surface type (such as plane, cylindrical surface, conical surface, etc.), and edges are encoded by contact / transition attributes between adjacent surfaces (such as concave / convex / smooth and direction combination encoding), and "non-adjacent" is used as an explicit missing edge constraint, thereby standardizing the "surface-edge adjacency relationship of groove" into a computable and comparable graph structure description.

[0130] Based on this description, this embodiment further encodes each slot template into a pre-matching subgraph matrix: the diagonal elements of the matrix correspond to the type encoding of each constituent face, and the off-diagonal elements correspond to the adjacency relationship encoding between faces (if adjacent) or -1 (if not adjacent). In addition, multiple compatibility template matrices are constructed for adjacency encoding variants that may occur in actual modeling (such as inner loop compatibility, direction reversal, smooth edge difference, etc.), forming a feature database that can be directly used for graph matching, providing structured input for subsequent efficient and robust feature recognition.

[0131] Furthermore, the pre-defined groove feature library contains a variety of typical groove processing feature types, covering closed grooves, arc-side closed grooves, bottomless closed grooves, arc-side bottomless closed grooves, simple wide-body grooves and their multiple directional compatible variants, bottomed open grooves and bottomless open grooves. Through matrix representation of surface type encoding (plane 1, cylindrical surface 2, conical arc surface 3) and edge adjacency encoding (convex edge 111, bottom convex edge 112, smooth edge 131, reverse convex edge 211, reverse smooth edge 231), a pre-matching subgraph matrix ranging from 5×5 to 3×3 is constructed, which can accurately characterize the topological differences of different groove types and handle common engineering problems such as inner loop compatibility, direction reversal and smooth edge variation in actual CAD models.

[0132] Specifically, such as Figure 5 The wide-body groove shown consists of three surfaces, including two side surfaces and one bottom surface, all of which are planar.

[0133] like Figure 6 The closed groove shown consists of five surfaces: two planar side surfaces, two cylindrical side surfaces, and a bottom surface, with the bottom surface being planar.

[0134] like Figure 7 The bottomless closed groove shown consists of four surfaces, including two planar side surfaces and two cylindrical side surfaces, and has no bottom surface structure.

[0135] like Figure 8 The bottomed open slot shown consists of four surfaces: two planar side surfaces, one cylindrical side surface, and one bottom surface, with the bottom surface being planar.

[0136] like Figure 9 The bottomless open slot shown consists of three surfaces: two planar sides and one cylindrical side, with no bottom surface.

[0137] S5. Based on the pre-matched subgraph matrix, candidate slot features that are consistent with the topology of each feature subgraph are selected from the attribute adjacency matrix of the model.

[0138] Specifically, such as Figure 10 As shown, step S5 includes:

[0139] S51. The pre-matching subgraph matrix corresponding to each slot feature template is taken as matrix A, and the adjacency graph extracted from the three-dimensional geometric model is taken as graph B.

[0140] S52. Convert the A matrix into an attribute adjacency graph structure, wherein its diagonal elements represent the surface type (plane 1, cylindrical surface 2, conical surface 3), the off-diagonal elements represent the adjacency relationship between surfaces (convex edge 111, smooth edge 131, etc.), and -1 indicates that they are not adjacent.

[0141] S53. Using a depth-first search algorithm, starting from the non-fragmented nodes in matrix A, search for candidate mapping results in graph B that satisfy the subgraph isomorphic mapping conditions. The subgraph isomorphic mapping conditions include:

[0142] 1) Node values ​​(face type) are exactly matched; 2) Edge weights (adjacency relationships) are completely consistent; 3) The mapping preserves all explicitly specified adjacency and non-adjacency constraints in matrix A;

[0143] S54. In the matching process, the impact of fragmented surface merging is handled. Through the mapping relationship between the representative surface and the original fragmented surface, multiple fragmented surfaces in the B diagram can be mapped to a single representative surface in the A matrix to ensure that fragmented surfaces do not affect the topology consistency judgment.

[0144] S55. After matching is completed, the candidate mapping results are normalized and cleaned, and invalid candidates containing duplicate face indices or face set conflicts are removed. A unique set of candidate slot feature instances is output as the candidate slot features.

[0145] Subsequently, candidate instances are constrained and verified according to the geometric rules corresponding to the slot type, and instances with geometrically invalid conditions are further eliminated, resulting in a high-quality slot feature set for subsequent deduplication and output. This method supports sequential matching and unified screening of multiple slot type templates and their compatible variants, and can adapt to changes in slot feature morphology caused by differences in boundary encoding and surface segmentation in engineering models, thereby improving the stability and accuracy of slot feature recognition.

[0146] S6. The candidate slot features are systematically filtered and categorized according to a preset order. Specifically, this includes:

[0147] Duplicate instances are removed according to recognition priority, and high-priority instances are retained to output the final slot feature set. Specifically, the recognition strategy of "strict first, then lenient, closed first, then open, and wide-body slot last" is adopted. Closed slot feature templates and arc-side closed slot feature templates are recognized first, followed by bottomless closed slot feature templates and arc-side bottomless closed slot feature templates, then open slot feature templates, and finally simple wide-body slot feature templates, so as to avoid the preemption of subsequent template candidate surfaces by the recognized instances.

[0148] S7. Based on the preset geometric rules of faces and edges, perform geometric constraint verification on the candidate slot features, eliminate redundant instances that do not meet the geometric constraints and have overlapping spatial positions, and obtain the final slot feature set.

[0149] Specifically, step S7 includes:

[0150] S71. For each type of slot feature template, obtain the candidate mapping results that satisfy the template topological constraints, and restore the set of faces involved in the candidate instances to the original set of faces based on the mapping relationship between the representative face and the original fragment face.

[0151] S72. Invoke the geometric rules corresponding to the slot type to perform geometric constraint verification. Among them, the geometric rules for closed slots include verification of the relative positional relationship between the bottom and the side, verification of the parallelism of the side, verification of the coaxiality of the cylindrical side, and verification of the radius consistency. The geometric rules for bottomless closed slots include performing cavity semantic judgment under the premise of missing bottom to exclude false features of boss type. The geometric rules for open slots include detection of the extension intersection of the side along the opening direction to exclude non-slot structures that intersect with the solid.

[0152] S73. After the geometric constraint verification is passed, deduplication is performed based on the set of occupied faces. If the new candidate instance has an intersection with the set of already retained instances, it is removed; otherwise, it is included in the final result and the set of occupied faces is updated.

[0153] S74. In the two-round recognition process of prioritizing the merged graph and supplementing the original graph, the consistent update logic of the occupied face set is maintained, so that the deduplication and retention decisions follow the preset order, thereby obtaining a final set of slot class feature instances that do not overlap and have consistent priority, which serves as the final slot class feature set.

[0154] It should be noted that for each type of slot feature, the system first obtains candidate mapping results that satisfy the template topological constraints in the model's adjacency graph, and then restores the face set involved in the candidate instance to the original face set (considering the mapping relationship between the representative face after merging the fragmented faces and the original fragmented faces). Subsequently, the system calls the geometric rules of the corresponding slot type for constraint verification: for example, for closed slots, the system verifies the relative relationship between the bottom and the side faces, the parallelism of the side faces, and the coaxiality and radius consistency of the cylindrical side faces; for bottomless slots, the system further performs cavity semantic judgment to exclude false features of the boss type under the premise of missing bottom faces; for open slots, the system excludes non-slot structures that intersect with the solid by detecting the extension intersection of the side faces along the opening direction. After the geometric verification is passed, the system uses the "occupied face set" as the deduplication criterion: if the new candidate instance and the retained instance have an intersection of face sets, the instance is directly removed; otherwise, it is included in the final result and the occupied face set is updated. Since this set is continuously accumulated throughout the process, the deduplication and retention decisions naturally follow the above recognition priority, and remain consistent between the two rounds of recognition of "merging graph priority, original graph supplementation", thus obtaining a set of slot feature instances that do not overlap and have consistent priority.

[0155] S8. Based on the final groove type feature set, output the groove processing features in the three-dimensional geometric model.

[0156] Reference Figure 11 As shown, after completing feature recognition and verification, the system outputs all common groove-type feature surfaces in the 3D geometric model, forming a structured set of feature information. This output process is based on candidate groove feature instances that have been geometrically constrained and verified. Each instance includes a feature type identifier, a set of constituent surface indexes, spatial positioning parameters, and a description of geometric attributes. For closed groove features, the system outputs a combination of surface indexes for two cylindrical side surfaces, two parallel plane side surfaces, and one plane bottom surface, along with geometric parameters such as the cylindrical surface radius, axial direction, and bottom surface normal vector. For arc-sided closed grooves, the system outputs the surface indexes for four cylindrical side surfaces and one bottom surface, emphasizing the coaxiality constraint and radius consistency of all cylindrical surfaces.

[0157] For bottomless groove features, output the set of side wall indexes and label the missing bottom features; for simple wide-body grooves, output the index combination of two parallel side surfaces and one bottom surface, and record the side surface spacing and bottom surface size information.

[0158] During the output process, a surface cluster reconstruction mechanism is employed. Through fragment mapping, discrete fragments in the actual model are mapped to representative surfaces in the feature template, ensuring that the output reflects both the original geometric composition and the semantic integrity of the features. All output feature surfaces undergo spatial relationship verification to avoid surface index conflicts or overlaps, guaranteeing that each physical surface belongs to only one dominant feature instance in the final feature set.

[0159] The final output feature surface set not only provides surface index information but also includes complete feature semantic descriptions and processing constraint information. This provides standardized and structured feature inputs for subsequent process planning, processing path generation, and manufacturing information management, realizing a digital mapping from geometric models to manufacturing features. This output system forms a closed-loop feature recognition-verification-output chain, ensuring the integrity, accuracy, and engineering applicability of common groove-type feature surfaces.

[0160] Example 2

[0161] Based on the method in Example 1, this example illustrates the implementation process of the method using a complex three-dimensional model containing various typical groove machining features. For example... Figure 12 The three-dimensional model shown. Specific methods include:

[0162] Step (1) extracts face-edge topology information from the 3D model, constructs a global adjacency graph of the model according to the face number, and forms the corresponding attribute adjacency matrix. Since the overall face-edge information is quite complex, for ease of explanation, only 10 nodes are selected here (…). Figure 13 The demonstration uses the medium-high gloss surface as an example, and its adjacency relationship is shown below. Figure 14 As shown.

[0163] Step (2) load various slot feature templates from the preset slot feature library, extract the face-edge topological relationships and corresponding attribute information of each face in the template, construct the corresponding feature subgraphs and convert them into feature subgraph matrices. The slot feature library includes seven types: simple wide-body slots, bottomed closed slots, bottomed circular arc closed slots, bottomless closed slots, bottomless circular arc closed slots, bottomed open slots, and bottomless open slots, as well as their compatible variants, as shown in Table 1.

[0164] Table 1: Common Groove Machining Feature Library

[0165] Common Threaded Hole Types Classification Number of nodes 1. Wide-body groove 3 2. Enclosed groove 5 3 Bottomless closed tank 4 4 Open slots 4 5 Bottomless Open Tank 3

[0166] Step (3) In response to the problem of fragmented surfaces that may be generated during the format conversion or modeling process of the 3D model, the model adjacency graph is preprocessed. By identifying cylindrical surface segments with a common axis and consistent radius, multiple fragmented surfaces are automatically merged and reconstructed into a single surface cluster node to obtain the surface cluster adjacency graph, so as to ensure the integrity and consistency of the topological structure of the slot feature.

[0167] Step (4) defines the matching criteria for nodes and directed edges on the adjacency graph of the face cluster, uses each feature subgraph matrix as a retrieval template, performs subgraph isomorphic matching operation in the adjacency matrix corresponding to the model face cluster adjacency graph, and initially selects a set of candidate slot feature surfaces with the same topology as the template.

[0168] Step (5) is to identify the seven types of groove features in the predefined order (bottomed closed groove → bottomed circular arc closed groove → bottomless closed groove → bottomless circular arc closed groove → bottomed open groove → bottomless open groove → simple wide-body groove). For each type of candidate groove feature surface, the preset geometric constraint rules (such as bottom flatness, sidewall parallelism, cylindrical surface coaxiality and related dimensional relationships) are applied to strictly verify and eliminate pseudo-feature instances that do not conform to the groove feature criteria of that type.

[0169] Step (6) involves deduplicating the slot feature instances that have passed geometric verification according to the above identification order. For candidate instances with obvious spatial overlap or face set inclusion relationships between different slot types and in different matching rounds, priority is given to retaining feature instances with higher priority and better geometric consistency, while removing redundant feature instances with lower priority. Finally, an effective set of slot features is collected. Figure 15 As shown, clicking on the identified groove features in the interface will highlight their corresponding groove feature surfaces; the final identification results are shown in Table 2, and the specific display effects of various groove features are as follows. Figures 16-22 As shown, the actual groove processing features in the model are accurately identified and clearly highlighted.

[0170] Table 2: Final Recognition Results

[0171] Recognizable categories quantity Wide body slot 1 Closed slot 3 Bottomless closed groove 0 Open slot 3 Bottomless Open Slot 0

[0172] Example 3

[0173] Based on the same inventive concept, this embodiment provides a groove processing feature recognition system. The principle of solving the problem is similar to that of the groove processing feature recognition method described above, and the repeated parts will not be repeated.

[0174] This embodiment provides a groove processing feature recognition system, including:

[0175] The 3D geometric model input module is used to input the 3D geometric model of a part with groove-like features;

[0176] The attribute adjacency matrix generation module is used to extract the edge adjacency relationships of each face in the three-dimensional geometric model, construct an adjacency graph, generate the attribute adjacency matrix of the model based on the adjacency graph, and identify and merge the fragmented faces caused by format conversion. The face patches with common axes and consistent geometric parameters are reconstructed into complete face clusters to ensure the continuity of the topological structure.

[0177] The slot database construction module is used to construct a slot database, which includes slot feature templates for: simple wide-body slots, bottom-closed slots, bottomless closed slots, bottom-open slots, and bottomless open slots.

[0178] The pre-matching subgraph matrix construction module is used to extract the face-edge adjacency relationship of each slot class feature template based on the slot class database and construct the corresponding pre-matching subgraph matrix; wherein, the face-edge adjacency relationship is represented by a matrix of face type encoding and edge adjacency encoding;

[0179] The candidate slot feature filtering module is used to filter out candidate slot features that are consistent with the topological structure of each feature subgraph from the attribute adjacency matrix of the model based on the pre-matched subgraph matrix.

[0180] An ordered filtering and classification module is used to perform ordered filtering and classification of the candidate slot features according to a preset order;

[0181] The final slot feature set acquisition module is used to perform geometric constraint verification on the candidate slot features according to the preset geometric rules of the face and edge, and remove redundant instances that do not meet the geometric constraints and have overlapping spatial positions to obtain the final slot feature set.

[0182] The groove machining feature output module is used to output the groove machining features in the three-dimensional geometric model based on the final groove type feature set.

[0183] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0184] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0185] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0186] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0187] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying groove processing features, characterized in that, include: Input a 3D geometric model of a part with groove-like features; The edge adjacency relationships of each face in the three-dimensional geometric model are extracted, an adjacency graph is constructed, and an attribute adjacency matrix of the model is generated based on the adjacency graph. The fragmented faces generated by the format conversion are identified and merged in the preprocessing. Faces with common axes and consistent geometric parameters are reconstructed into complete face clusters to ensure the continuity of the topological structure. Construct a slot class database, which includes slot class feature templates for: simple wide-body slots, bottom-closed slots, bottomless closed slots, bottom-open slots, and bottomless open slots; Based on the slot database, the face-edge adjacency relationships of each slot feature template are extracted to construct the corresponding pre-matching subgraph matrix; wherein, the face-edge adjacency relationships are represented by matrix representation of face type encoding and edge adjacency encoding; Based on the pre-matched subgraph matrix, candidate slot features that are consistent with the topology of each feature subgraph are selected from the attribute adjacency matrix of the model. The candidate slot features are systematically filtered and categorized according to a preset order; Based on the preset geometric rules of faces and edges, the candidate slot features are geometrically constrained and verified. Redundant instances that do not meet the geometric constraints and overlap in spatial position are eliminated to obtain the final slot feature set. Based on the final groove feature set, the groove machining features in the three-dimensional geometric model are output.

2. The method for identifying groove processing features according to claim 1, characterized in that, Extract the edge adjacency relationships of each face in the three-dimensional geometric model and construct an adjacency graph, including: Each face of the three-dimensional geometric model is labeled with a face ID, and adjacency graph nodes are created using the face IDs. For any shared edge in the three-dimensional geometric model, establish a bidirectional topological association connecting the two sides of the edge to form an adjacency graph containing directed edges and adjacency graph nodes.

3. The method for identifying groove processing features according to claim 2, characterized in that, Based on the adjacency graph, an attribute adjacency matrix of the model is generated, including: Encode the face attributes into the diagonal elements of the attribute adjacency matrix, wherein the face attributes include face geometry type, normal vector, and area; Edge attributes are encoded into the off-diagonal elements of the attribute adjacency matrix, and the edge attributes include length, concavity and smoothness, inner and outer loop markings, and edge geometry type.

4. The method for identifying groove processing features according to claim 3, characterized in that, The diagonal elements are encoded using a surface type, and the five primitive types are defined as follows: 1 for plane, 2 for cylinder, 3 for cone, 4 for sphere, and 5 for torus. The off-diagonal elements are encoded using adjacency codes, and the encoding rules are as follows: code = X×100 + Y×10 + Z; Where X∈{1,2}, 1 and 2 represent the outer ring and the inner ring respectively; Y∈{1,2,3}, where 1, 2, and 3 represent concave edge, convex edge, and smooth edge, respectively. Z∈{1,2,3,4}, where 1, 2, 3, and 4 represent the sides as a straight line, a circle, a closed curve, and an ellipse, respectively.

5. The method for identifying groove processing features according to claim 1, characterized in that, Fragmented surfaces generated by format conversion are identified and merged in a preprocessing step. Surfaces with a common axis and consistent geometric parameters are reconstructed into complete surface families to ensure the continuity of the topology. This includes: After constructing the adjacency graph, while maintaining the face adjacency relationship, the nodes that are coaxial and have the same diameter, belong to the same cylindrical analytical surface but are split into multiple fragments are clustered into the same face cluster, based on the consistency of the cylindrical geometric parameters. Subsequently, the graph is folded and reconstructed using face clusters as the basic unit. Representative face nodes are selected within the face cluster, and the external adjacent edges of each face within the face cluster are merged. Edges pointing to any member within the cluster are uniformly redirected to the representative node, thereby constructing a face cluster adjacency graph with node semantics of "complete cylindrical face" and edge semantics inheriting the original adjacency attributes.

6. The method for identifying groove processing features according to claim 1, characterized in that, Based on the slot class database, the face-edge adjacency relationships of each slot class feature template are extracted, and the corresponding pre-matching subgraph matrix is ​​constructed, including: Starting from the structural definition of the feature templates of each type of groove, each type of groove is decomposed into several component surfaces with clear processing semantics. The component surfaces include the bottom surface, two side surfaces, end arc surface or end conical arc surface. The face-edge topological adjacency relationships between the constituent faces are extracted as the edge set of the template subgraph. The nodes are encoded using face type, which includes plane, cylindrical, and conical surfaces. The edges are encoded using edge adjacency encoding to encode the contact or transition attributes between adjacent faces. The contact or transition attributes include concave edges, convex edges, smooth edges, and direction combination information. By treating non-adjacent relationships as explicit missing edge constraints, the face-edge adjacency relationships of the slot-type features are normalized into a computable and comparable graph structure description. Based on the graph structure description, each slot template is encoded as the pre-matching subgraph matrix, wherein the diagonal elements of the pre-matching subgraph matrix are the face type codes of each constituent face, and the off-diagonal elements of the pre-matching subgraph matrix are the face edge adjacency codes. When two constituent faces are not adjacent, the corresponding element is -1. To address the edge adjacency encoding variants that occur in actual modeling, including inner loop label compatibility, adjacency direction reversal, and smooth edge difference, multiple compatibility template pre-matching subgraph matrices are constructed and written into the slot class database to form a feature database for subsequent graph matching.

7. The method for identifying groove processing features according to claim 1, characterized in that, Based on the pre-matched subgraph matrix, candidate slot features consistent with the topological structure of each feature subgraph are selected from the attribute adjacency matrix of the model, including: The pre-matched subgraph matrix corresponding to each slot feature template is taken as matrix A, and the adjacency graph extracted from the three-dimensional geometric model is taken as graph B. The matrix A is converted into an attribute adjacency graph structure, where the diagonal elements represent the face types and the off-diagonal elements represent the adjacency relationships between faces. A depth-first search algorithm is used to search for candidate mapping results in graph B that satisfy the subgraph isomorphic mapping conditions, starting from the non-fragmented nodes in matrix A. The subgraph isomorphic mapping conditions include: exact matching of node values, completely consistent edge weights, and maintaining the explicitly specified adjacency and non-adjacency constraints in matrix A. The impact of fragmented surface merging is addressed during the matching process. By mapping the representative surface to the original fragmented surface, multiple fragmented surfaces in the B diagram can be mapped to a single representative surface in the A matrix, ensuring that fragmented surfaces do not affect the topology consistency judgment. After matching is completed, the candidate mapping results are normalized and cleaned, invalid candidates containing duplicate face indices or face set conflicts are removed, and a unique set of candidate slot feature instances is output as the candidate slot features.

8. The method for identifying groove processing features according to claim 1, characterized in that, The candidate slot features are systematically filtered and categorized according to a preset order, including: Prioritize identifying closed groove feature templates and arc-side closed groove feature templates, then identify bottomless closed groove feature templates and arc-side bottomless closed groove feature templates, then identify open groove feature templates, and finally identify simple wide-body groove feature templates.

9. The method for identifying groove processing features according to claim 1, characterized in that, Based on preset geometric rules for faces and edges, the candidate slot features are geometrically constrained and verified, including: For each type of slot feature template, obtain the candidate mapping results that satisfy the template topological constraints, and restore the set of faces involved in the candidate instances to the original set of faces based on the mapping relationship between the representative face and the original fragment face; The geometric rules corresponding to the slot type are invoked for geometric constraint verification. Among them, the geometric rules for closed slots include verification of the relative positional relationship between the bottom and the side, verification of the parallelism of the side, verification of the coaxiality of the cylindrical side, and verification of the radius consistency. The geometric rules for bottomless closed slots include performing cavity semantic judgment to exclude false features of the boss type when the bottom is missing. The geometric rules for open slots include detecting the extension intersection of the side along the opening direction to exclude non-slot structures that intersect with the solid. After the geometric constraint verification is passed, deduplication is performed based on the set of occupied faces. If a new candidate instance has an intersection with the face set of a retained instance, it is removed; otherwise, it is included in the final result and the set of occupied faces is updated. In the two-round recognition process of prioritizing merging the graph and supplementing the original graph, the consistent update logic of the occupied face set is maintained, so that the deduplication and retention decisions follow the preset order, thereby obtaining a final set of slot class feature instances that do not overlap and have consistent priority, which serves as the final slot class feature set.

10. A groove-type machining feature recognition system, characterized in that, include: The 3D geometric model input module is used to input the 3D geometric model of a part with groove-like features; The attribute adjacency matrix generation module is used to extract the edge adjacency relationships of each face in the three-dimensional geometric model, construct an adjacency graph, generate the attribute adjacency matrix of the model based on the adjacency graph, and identify and merge the fragmented faces caused by format conversion. The face patches with common axes and consistent geometric parameters are reconstructed into complete face clusters to ensure the continuity of the topological structure. The slot database construction module is used to construct a slot database, which includes slot feature templates for: simple wide-body slots, bottom-closed slots, bottomless closed slots, bottom-open slots, and bottomless open slots. The pre-matching subgraph matrix construction module is used to extract the face-edge adjacency relationship of each slot class feature template based on the slot class database and construct the corresponding pre-matching subgraph matrix; wherein, the face-edge adjacency relationship is represented by a matrix of face type encoding and edge adjacency encoding; The candidate slot feature filtering module is used to filter out candidate slot features that are consistent with the topological structure of each feature subgraph from the attribute adjacency matrix of the model based on the pre-matched subgraph matrix. An ordered filtering and classification module is used to perform ordered filtering and classification of the candidate slot features according to a preset order; The final slot feature set acquisition module is used to perform geometric constraint verification on the candidate slot features according to the preset geometric rules of the face and edge, and remove redundant instances that do not meet the geometric constraints and have overlapping spatial positions to obtain the final slot feature set. The groove machining feature output module is used to output the groove machining features in the three-dimensional geometric model based on the final groove type feature set.

Images