摘要
飞机装配中产生的间隙阶差都有具体要求,间隙阶差的提取精度保证了装配的质量。针对因缝隙的尺寸不均匀、数据采集存在噪声导致难以精确提取间隙阶差的问题,提出了一种基于支持向量机(Support Vector Machine,SVM)的三维对缝点云间隙阶差提取方法。首先根据对缝点云分布特点,建立间隙阶差数学模型,明确所需提取的特征点;其次根据数模的边界进行测量点规划与离散,以测量点为几何中心,利用主成分分析(Principal Component Analysis,PCA)与包围盒法提取出子点云;接着调整SVM超平面,分割点云;然后对点云三角划分,根据单边准则提取点云边界点,根据边界点、超平面的几何关系提取边缘点与临界点;最后根据间隙阶差的数学模型,提取间隙阶差值,并设计试验验证了该算法的精度与稳定性。试验表明:该方法的间隙测量均值误差在0.03mm以下,阶差测量均值误差在0.02mm以下。
Gap and flush generated among aircraft assembly should satisfy specific requirements,which should be precisely extracted in order to ensure the quality of assembly.Because of seam’s uneven size and noise of the data,gap and flush are difficult to extract accurately.Therefore,this paper proposes a method to extract gap and flush of threedimensional point cloud of seam based on SVM(Support Vector Machine).Firstly,mathematical model is established according to distribution characteristics of point cloud,in order to identify the feature points that need to be extracted.Secondly,measuring position points are planned and discreted according to the boundary of the digital model.Take the measured position points as the geometric center,and subpoint cloud is extracted based on PCA(Principal Component Analysis)and bounding box.Thirdly,point cloud is segmented by hyperplane which is adjusted.Then,boundary points,edge points and critical points are extracted by triangler point cloud.Finally,gap and flush are extracted according to mathematical models.The accuracy and stability of the method were verified through designed experiments.Experiments show that gap measurement mean error is less than 0.03mm as well as the flush measurement mean error is less than 0.02mm.
作者
张波
李泷杲
郝龙
主逵
ZHANG Bo;LI Shuanggao;HAO Long;ZHU Kui(College of Mechanical and Electrical Engineering,Nanjing University of Aeronautics&Astronautics,Nanjing 210016,China;Shenzhen JT Automation Equipment Co.,Ltd.,Shenzhen 518216,China)
出处
《航空制造技术》
2020年第7期47-54,共8页
Aeronautical Manufacturing Technology
关键词
点云分割
支持向量机(SVM)
特征点
间隙阶差
主成分分析(PCA)
Point cloud segmentation
Support Vector Machine(SVM)
Feature points
Gap and flush
Principal Component Analysis(PCA)
