数据驱动的飞机结构件加工特征识别方法  

A Data-Driven Method for Machining Feature Recognition for Aircraft Structural Parts

作　　者：陆凯 李迎光[1] 刘旭 邓天池 LU Kai;LI Yingguang;LIU Xu;DENG Tianchi(College of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;School of Mechanical and Power Engineering,Nanjing Tech University,Nanjing 210009,China)

机构地区：[1]南京航空航天大学机电学院,南京210016 [2]南京工业大学机械与动力工程学院,南京210009

出　　处：《航空制造技术》2021年第8期32-40,57,共10页Aeronautical Manufacturing Technology

基　　金：国家杰出青年科学基金(51925505)。

摘　　要：加工特征识别技术是实现CAD/CAM一体化的重要途经,对提升飞机结构件数控加工质量和效率具有重要意义。飞机结构件结构复杂,包含大量复杂相交特征,现有加工特征识别方法识别相交特征时需要预先定义特征相交模式,难以满足飞机结构件复杂相交特征识别需求。因此,提出了数据驱动的飞机结构件加工特征识别方法,将加工特征识别问题转化为图学习问题,通过构建图神经网络模型从历史工艺数据中自适应学出特征识别规则,打破了特征相交模式预先定义思路的局限。以典型飞机结构件为测试对象,孤立特征、相交特征和总的加工特征正确识别率分别达到了98.11%、94.62%和96.18%,验证了所提方法的有效性。Machining feature recognition is an essential way to realize the integration of CAD/CAM,and is significant to improve the quality and efficiency of CNC machining of the aircraft structural parts,which contains plenty of complex interacting features with complex structure.Existing machining feature recognition methods require predefinition of feature interacting patterns before recognizing interacting features,which is difficult to meet the recognition requirements of complex interacting features of the aircraft structural parts.Therefore,a data-driven method for machining feature recognition for aircraft structural parts is proposed,which transforms the problem of machining feature recognition into a graph learning problem,and adaptively learns feature recognition rules from historical process data by constructing a graph neural network model.The proposed method breaks the limits of the pre-defined feature interacting patterns.With typical aircraft structural parts as test parts,the correct recognition rates of isolated features,interacting features and total machining features reach 98.11%,94.62%and 96.18%,respectively,which verifies the effectiveness of the proposed method.

关 键 词：数据驱动 飞机结构件 加工特征识别 相交特征 图神经网络 

分 类 号：V262[航空宇航科学与技术—航空宇航制造工程]