计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (10): 156-163.doi: 10.11896/jsjkx.250800066

• 计算机图形学 & 多媒体 • 上一篇    下一篇

基于图Transformer的CAD模型加工特征识别方法

蓝旺, 周宇, 贺志飞   

  1. 南京航空航天大学计算机科学与技术学院 南京 210016
  • 收稿日期:2025-08-18 修回日期:2025-12-26 发布日期:2026-10-13
  • 通讯作者: 周宇(zhouyu@nuaa.edu.cn)
  • 作者简介:(wanglan@nuaa.edu.cn)
  • 基金资助:
    国防基础科研项目(JCKY2022605C006)

CAD Model Machining Feature Recognition Method Based on Graph Transformer

LAN Wang, ZHOU Yu, HE Zhifei   

  1. College of Computer Science and Technology,Nanjing University of Aeronautics and Astronautics,Nanjing 210016, China
  • Received:2025-08-18 Revised:2025-12-26 Online:2026-10-13
  • About author:LAN Wang,born in 1999,postgraduate.His main research interest is 3D model feature recognition.
    ZHOU Yu,born in 1981,Ph.D,graduate advisor.His main research interest is intelligent software development.
  • Supported by:
    National Defense Basic Scientific Research Program(JCKY2022605C006).

PDF (PC)

86

摘要： 加工特征识别在计算机辅助设计(CAD)和制造(CAM)中至关重要,是连接CAD和CAM的关键。传统的加工特征识别方法主要有基于规则和基于学习的方法,但是这些方法存在泛化性差、维护成本高等问题。针对上述问题,提出了一种基于图Transformer的CAD模型加工特征识别方法,该方法以边界表示法(B-Rep)描述 CAD 模型的几何拓扑结构,通过面邻接图建模三维模型的局部与全局空间关系,用多通道卷积神经网络(CNN)对曲面和曲线的几何属性进行离散化编码,再利用图 Transformer多头注意力机制动态捕捉节点间的长距离依赖与交互特征,最终通过多层感知机(MLP)实现加工特征的分类。在公开数据集MFCAD++上的实验结果表明,该方法取得了99.73%的分类准确率(Acc)、99.65%的平均类别准确率(Per-class Acc)和99.58%的平均交并比(mIoU)。此外,消融实验验证了扩展邻接矩阵与属性编码等关键模块对模型鲁棒性的贡献,而域自适应方法的引入有效提升了模型的泛化能力。

关键词: 加工特征识别, B-Rep表示法, 图Transformer, 多层感知机, 扩展邻接矩阵

Abstract: Machining feature recognition is of vital significance in computer-aided design(CAD) and computer-aided manufactu-ring(CAM),acting as a critical link that connects CAD and CAM systems.Traditional machining feature recognition methods are primarily categorized into rule-based and learning-based approaches.However,these methods exhibit inherent limitations such as poor generalization capability and high maintenance costs.To address the aforementioned issues,this paper proposes a graph Transformer-based method for machining feature recognition of CAD models.Specifically,the boundary representation(B-Rep) is adopted to describe the geometric and topological structure of CAD models.Face adjacency graphs are utilized to model the local and global spatial relationships of 3D models,while a multi-channel convolutional neural network(CNN) is employed to perform discrete encoding on the geometric attributes of surfaces and curves.Furthermore,the multi-head self-attention mechanism of the graph Transformer is leveraged to dynamically capture long-range dependencies and interactive features between nodes.Finally,a multi-layer perceptron(MLP) is used to realize the classification of machining features.On the public MFCAD++ dataset,the proposed method exhibits exceptional performance,reaching 99.73% in classification accuracy(Acc),99.65% in per-class accuracy,and 99.58% in mean intersection over union(mIoU).Ablation experiments confirm the crucial role of components like the extended adjacency matrix and attribute encoding in ensuring model robustness.Moreover,the integration of domain adaptation significantly improves the model's generalization ability.

Key words: Machining feature recognition, B-Rep representation, Graph Transformer, Multi-layer perceptron, Extended adjacency matrix

中图分类号: 

  • TP391
[1] HE H C,FANG Z,LU Z D,et al.PT-MFR:a CAD model machining feature recognition method based on Point Transformer[J].Journal of University of Chinese Academy of Sciences,2024,43(5):3-8.
[2] NING F,SHI Y,CAI M,et al.Part machining feature recognition based on a deep learning method[J].Journal of Intelligent Manufacturing,2023,34(2):809-821.
[3] ANGRISH A,CRAVER B,STARLY B.“FabSearch”:A 3D CAD model-based search engine for sourcing manufacturing services[J].Journal of Computing and Information Science in Engineering,2019,19(4):041006.
[4] ZHANG Z,JAISWAL P,RAI R.Featurenet:Machining feature recognition based on 3d convolution neural network[J].Computer-Aided Design,2018,101:12-22.
[5] WORNER J M,BROVKINA D,RIEDEL O.Feature recognition for graph-based assembly product representation using machine learning[C]//2021 21st International Conference on Control,Automation and Systems(ICCAS).IEEE,2021:629-635.
[6] HE Y,YU H,LIU X,et al.Deep learning based 3D segmentation:A survey[J].arXiv:2103.05423,2021.
[7] YEO C,KIM B C,CHEON S,et al.Machini-ng feature recognition based on deep neural networks to support tight integration with 3D CAD systems[J].Scientific Reports,2021,11(1):22147.
[8] MATURANA D,SCHERER S.Voxnet:A 3d convolutionalneural network for real-time object re-cognition[C]//2015 IEEE/RSJ International Conference on Intelligent Robots and Systems(IROS).IEEE,2015:922-928.
[9] JIA L,ZHANG S W,CAO Y R,et al.Machining feature recognition method based on improved mesh neural network[J].Iranian Journal of Science and Technology,Transactions of Mechanical Engineering,2023,47(4):2045-2058.
[10] JIAN C,SHENG Z,LU Y,et al.a quaternion se-mantic cell convolution graph neural network for MBD product model classification[J].IEEE Transactions on Industrial Informatics, 2023,19(12):11477-11486.
[11] WANG Q M,YU X H.Ontology based automatic feature recognition framework[J].Computers in Industry,2014,65(7):1041-1052.
[12] JOSHI S,CHANG T C.Graph-based heuristics for recognition of machined features from a 3D solid model[J].Computer Aided Des,1988,20(2):58-66.
[13] GAO S,SHAH J J.Automatic recognition of interacting machining features based on mini-mal condition subgraph[J].Computer Aided Design,1998,30(9):727-739.
[14] YUEN C,VENUVINOD P K.Geometric feature recognition:coping with the complexity and infinite variety of features[J].International Journal of Computer Integrated Manufacturing,1999,12(5):439-452.
[15] VANDENBRANDE J H,REQUICHA AA G.Spatial reasoning for the automatic recognition of machinable features in solid models[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1993,15(12):1269-1285.
[16] ZEHTABAN L,ROLLER D.Automated rule-based system for Opitz feature recognition and code generation from STEP[J].Computer-Aided Design and Applications,2016,13(3):309-319.
[17] HUANG Z,YIP-HOI D.High-level feature recognition usingfeature relationship graphs[J].Computer-Aided Design,2002,34(7):561-582.
[18] GENG W,CHEN Z,HE K,et al.Feature recognition and volume generation of uncut regions for electrical discharge machining[J].Advances in Engineering Software,2016,91:51-62.
[19] WU H,LEI R,PENG Y,et al.AAGNet:A graph neural network towards multi-task machining feature recognition[J].Robotics and Computer-Integrated Manufacturing,2024,86:102661.
[20] LEI R,WU H,PENG Y.MFPointNet:A point cloud-based neural network using selective downsampling layer for machining feature recognition[J].Machines,2022,10(12):1165.
[21] PEDDIREDDY D,FU X,SHANKAR A,et al.Identifying manufacturability and machining processes using deep 3D convolutional networks[J].Journal of Manufacturing Processes,2021,64:1336-1348.
[22] JIA J L,ZHANG S W,CAO Y R,et al.Machining feature recognition method based on improved mesh neural network[J].Iranian Journal of Science and Technology,Transactions of Mechanical Engineering,2023,47:2045-2058
[23] SHI P,QI Q,QIN Y,et al.Highly interacting machining feature recognition via small sample learning[J].Robotics and Compu-ter-Integrated Manufacturing,2022,73:102260.
[24] CAO W,ROBINSON T,HUA Y,et al.Graphrepresentation of 3D CAD models for machining feature recognition with deep learning[J].American Society of Mechanical Engineers 2022,31(4):811-821.
[25] COLLIGAN A R,ROBINSON T T,NOLAN D C,et al.Hierarchical CADNet:Learning from B-reps for machining feature re-cognition[J].Computer-Aided Design,2022,147:103226.
[26] SHI P Z,QI Q F,QIN Y C,et al.Intersecting machining feature localization and recognition via single shotmultibox detector[J].IEEE Transactions on Industrial Informatics,2021,17(5):3292-3302.
[27] YAO X H,WANG D,YU T,et al.A machining feature recognition approach based on hierarchical neural network for multi-feature point cloud models[J].Journal of Intelligent Manufacturing,2023,34(6):2599-2610.
[28] QI C R,YI L,SU H,et al.PointNet++:Deep hierarchical feature learning on point sets in a metric space[J].Advances in neural information processing systems,2017,30:5105-5114.
[29] GONG Q,JIANG J F,CHEN Z M,et al.Global Symmetry Detection for B-Rep Models Based on Partial-Prior Recognition[J].Journal of Computer-Aided Design & Computer Graphics,2017,29(2):384-392.
[30] JAYARAMAN P K,SANGHI A,LAMBOURNE J G,et al.UV-Net:Learning from boundary representations[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).IEEE,2021:11698-11707.
[31] QIAN X Y,SHI Y Z,ZHANG F,et al.Transformer Object Tracking Based on Re-Parameterization Network[J].Journal of Computer-Aided Design & Computer Graphics,2023,35(10):1521-1531.
[32] ZHANG S,GUAN Z,JIANG H,et al.BrepMFR:Enhancingmachining feature recognition in B-rep models through deep learning and domain adaptation[J].Computer Aided Geo-metric Design,2024,111:102318.
[33] CHA M H,KIM B C.Machining feature recognition using B-RepNet[J].Journal of Mechanical Science and Technology,2023,37(12):6103-6113.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发