Research On Key Technique Of Point Cloud Processing For Semi-Physical Model Of Large Size Aerospace Structure Assembly

With the rapid development of economy and improvement of people's living standards,the manual fabric such as carpet and embroidery has been widely used in households,hotels,exhibition halls,companies and other places,which makes it of high demand and a broad market prospect.Large-scale aerospace structural parts manufacturing bias is inevitable,when there is processing error on the workpiece,it is easy to lead to assembly problems.Before the actual installation,simulate and verify the actual assembly results by using three-dimensional measurement technology,allocate tolerance through human-computer interaction,will achieve onetime assembly success,improve assembly efficiency and control assembly accuracy effectively.In this paper,with semi-physical model point cloud processing problems for large-scale aerospace structural parts,works such as sampling of high density point cloud data,point cloud reduction,point cloud denoising,point cloud feature extraction,point cloud region segmentation,registration of point cloud data and theoretical CAD model,structural part processing and assembly deviation calculation were studied.The main work is as follows:(1)Based on the semi-physical virtual assembly model,a method of theoretical CAD model and high-density measurement data fusion is studied.In this paper,a semi-physical virtual assembly model is studied,and a semi-physical model of theoretical CAD model and mass cloud data interconnection is established,the storage of point cloud data and geometric model in semi-physical model are studied.(2)For the problem of inefficiency that the traditional ICP algorithm searches corresponding points,an improved nearest point search method is proposed.The method is based on region mapping to search the corresponding point,which can greatly reduce the scanning point cloud to the theoretical CAD Model of the nearest point of the search area,which is a good solution to time-consuming problem the traditional ICP algorithm to find the most recent points,greatly reducing the time required for point cloud and CAD model matching.(3)The traditional ICP registration algorithm only considers the fitting of geometric elements,but the semi-physical assembly model not only requires geometric fit,but also preferentially ensures the registration accuracy of the assembly characteristics.An improved ICP algorithm based on assembly feature weighting factor is proposed.The algorithm takes into account the constraint effect of the assembly feature in the registration process and assigns the weighting factor to the assembly feature.Experiments show that the algorithm can avoid the traditional ICP algorithm easily lead to local optimization,and improve the registration speed and registration accuracy.(4)Finally,a three-dimensional laser scanning system with large-size structural parts and a software system based on semi-physical model automatic measurement and point cloud processing are developed for the practical application of semi-physical assembly of spacecraft.The paper uses the developed scanning system to realize the accurate extraction of the cloud data of the equipment cabin.At the same time,it completes the functions of organization,management,streamlining,denoising,feature extraction and regional segmentation of point cloud data,and carries out point cloud data and theoretical CAD registration,calculate the processing error of the cabin,and visualization of the color cloud.The experimental results show that the proposed method can effectively deal with the assembly problem of large-scale aerospace structures based on the point cloud processing method based on the semi-physical assembly model.The proposed ICP algorithm based on the assembly feature weighting factor can effectively shorten the nearest point search time and reduce the registration iteration Times,thus improving the matching efficiency,and at the same time can improve the registration accuracy.