A Data-Driven Method for Machining Feature Recognition for Aircraft Structural Parts
LU Kai1 , LI Yingguang1 , LIU Xu2 , DENG Tianchi1
(1. College of Mechanical and Electrical Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China; 2. School of Mechanical and Power Engineering, Nanjing Tech University, Nanjing 210009, China)
Abstract: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.