基于特征分离的飞机结构件特征识别技术研究  

Research on Feature Recognition Technology of Aircraft Structural Parts Based on Process Feature Separation

作　　者：刘振宇 曹珍珍 杨青平 茅健 张立强 LIU Zhenyu;CAO Zhenzhen;YANG Qingping;MAO Jian;ZHANG Liqiang(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201600;Technology Center,Chengdu Yongfeng Technology Co.,Ltd.,Chengdu 610000)

机构地区：[1]上海工程技术大学机械与汽车工程学院,上海201600 [2]成都永峰科技有限公司技术中心,成都610000

出　　处：《计算机与数字工程》2023年第10期2452-2456,共5页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:51775328)资助。

摘　　要：为了有效识别并分离出飞机结构件的加工特征,论文通过定义飞机结构件的加工特征面,利用CAA二次开发技术,将加工特征定义为加工特征与辅助特征两种类型。利用邻接矩阵面节点间连接弧的凹凸性定义加工特征与辅助特征,并进行分离;将加工特征离散为点云数据,并采用DBSCAN算法对加工特征进行区分,得到单一特征与相交特征点云数据;输入点云数据至DGCNN模型的分类分割网络,实现单一特征的分类与相交特征的分割。开发CATIA应用插件对该方法进行验证,结果表明该方法能够准确识别出结构件中的复杂特征。In order to effectively identify and separate the process features of aircraft structural parts,this paper defines the process features of aircraft structural parts into two types,processing features and auxiliary features,by defining the processing features of aircraft structural parts and using CAA secondary development technology.The machining features and auxiliary features are defined and separated by the concavity of the connecting arcs between the adjacent matrix surface nodes.The processing features are separated into point cloud data,and the DBSCAN algorithm is used to separate the processing features,and the single feature and intersection feature point cloud data are obtained.The point cloud data is input into the classification and segmentation network of DGCNN model to achieve the classification of single features and segmentation of intersecting features.The CATIA application plug-in is developed to verify the method,and the results show that the method can accurately identify complex features in structural parts.

关 键 词：飞机结构件 CAA二次开发 特征分离 DBSCAN算法 DGCNN算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]