基于定位工艺几何特征的机械零件数字化检测方法研究

发布时间：2018-05-24 22:29

本文选题：数字化检测 + 点云　；参考：《南京航空航天大学》2017年硕士论文

【摘要】：近年来,三维点云获取手段得到迅猛发展,以点云为基础的三维数字化检测技术成为研究热点。三维数字化检测技术关键在于点云数据预处理、配准算法以及误差分析显示等。机械零件在实际加工制造中拥有高精度定位工艺几何特征,本文引入该特征作为点云与CAD模型配准的约束基准,进行机械零件检测方法研究,设计开发基于定位工艺几何特征的机械零件数字化检测系统,提高零件数字化检测精度。本文对基于定位工艺几何特征的机械零件数字化检测方法进行研究,包括点云预处理以及特征配准算法的研究,采用八叉树建立几何拓扑信息,提高了散乱点云数据的处理速度;通过采样算法,降低了点云冗余度;使用改进Z-score模型对点云数据去噪,降低噪声对测量数据的影响,提高点云数据准确性;提出了基于模型驱动的点云数据分割算法,将点云数据按照CAD的曲面信息进行分割,为特征创建以及特征约束配准奠定基础。对六种常见特征的拟合方法进行了研究,创建高精度的特征对;基于特征约束属性,设计出特征约束配准算法,使用高精度特征进行坐标配准,提高了配准精度;对误差计算方法以及可视化技术进行研究,开发出了基于定位工艺几何特征的机械零件数字化检测系统。通过对基于定位工艺几何特征的机械零件数字化检测方法研究,充分利用了高精度的定位工艺几何特征,提高了三维数字化检测的精度,对应用于实际工业形变检测分析具有重要意义。
[Abstract]:In recent years, 3D point cloud acquisition has been developed rapidly, and 3D digital detection technology based on point cloud has become a research hotspot. The key technologies of 3D digital detection are point cloud data preprocessing, registration algorithm and error analysis and display. Mechanical parts have the geometric feature of high precision positioning technology in actual machining and manufacture. This paper introduces this feature as the constraint datum of point cloud matching with CAD model, and carries on the research on the inspection method of mechanical parts. A digital testing system for mechanical parts based on geometric features of positioning process is designed and developed to improve the accuracy of digital detection of mechanical parts. In this paper, the digital detection method of mechanical parts based on the geometric feature of positioning technology is studied, including the point cloud preprocessing and the algorithm of feature registration, and the geometric topology information is established by octree. The processing speed of scattered point cloud data is improved, the redundancy of point cloud is reduced by sampling algorithm, the point cloud data is de-noised by using improved Z-score model, the influence of noise on measurement data is reduced, and the veracity of point cloud data is improved. A model-driven point cloud data segmentation algorithm is proposed. The point cloud data is segmented according to the surface information of CAD, which lays the foundation for feature creation and feature constraint registration. The fitting methods of six common features are studied, and high precision feature pairs are created. Based on the feature constraint attribute, the feature constraint registration algorithm is designed, and the coordinate registration with high precision features is used to improve the registration accuracy. Based on the research of error calculation method and visualization technology, a digital inspection system for mechanical parts based on geometric feature of positioning process is developed. Through the research on the digital detection method of mechanical parts based on the geometric feature of positioning technology, the high precision geometric feature of positioning process is fully utilized, and the accuracy of 3D digital detection is improved. It is of great significance for practical industrial deformation detection and analysis.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG80

【参考文献】