| The freeform surface not only has a good shape performance but also exhibits excellent mechanical properties.It has been increasingly applied to aerospace,vehicle and shipbuilding,machinery and die and other fileds.At the same time,the free-form surface parts with complex shapes have great difficulties in the surfacing welding of repairing automatically,not only in the weld tracking,the timely adjustment of the welding torch deviation,especially,in the recognition of the surfacing surface.At present,the surfacing repair of complex shaped parts is mainly manual.The harsh environment not only seriously damages people’s health,but also hardly guarantees a good welding quality of artificial surfacing.Simultaneously,with the advancement of industrial technology,the size of parts has become even greater,the work of manual surfacing get more cumbersome.There is an urgent need for a new welding technique for automated surfacing free-form surfaces.The paper mainly studies the rotating arc automatic surfacing free-form.The main idea of the article is to recognition the free surface by analyzing the sampling point cloud.First of all,a sampling point relationship model is established to describe the distribution rule of the sample point in the surface,and seek the position coordinates of the sample point cloud in the spatial absolute coordinate system;Then,the scattered point cloud is regularized,the feature curve is identified,and the deviation information is extracted.At last,the surface to be welded is recognized by surface patch fitting.The specific research contents are as follows:Firstly,the free surfacing model based on rotating arc is established,and the model to describe the free surface sample principle is designed.Through the free surface sampling point relationship model,the position of each point in the absolute coordinate system is determined,and the coordinate relationship between each sampling point is given.Secondly,the characteristics of the sampling point cloud are analyzed,the point cloud characteristic line and the welding seam deviation are extracted.,and the trans-formation relationship between the scattered sampling point cloud and the regularized point cloud is established in the base of the discrete sampling point cloud.The princi-pal component analysis method combines with the features of feature discontinuities at the projected points of the principal plane to extract feature points.Finally,the point cloud characteristic curve and weld misalignment letter are extracted in the base of feature points are filtered by feature point constraint functions.Then,the topological relationship between scattered sample points is established and stored in the space cube box,as an octree.The point cloud standard deviation threshold and point cloud segmentation depth are used as the basis for the point cloud segmentation to cluster the points in the scattered sampling point cloud.The whole point cloud is divided into different point cloud subsets with different characteristics,and the point cloud density is better in the subcube box,and the accurate and smooth surface patch fitting of the point cloud in the subset is achieved.Finally,a surface recognition algorithm is proposed to reconstruct the curved sur-face using surface mosaics.The surface equations at the patch mosaic are searched,and the differential operator algorithm is used to replace the idea of smoothing tran-sitions at the joints by the derivative operator to complete the surface reconstruction.In combination with the above methods for seam tracking and surface reconstru-ction,experimental data are used to analyze the algorithm through simulation.The maximum left-side error of the weld is 0.36mm in the recognition result,the maximu-m deviation in the right direction of the weld is 0.24mm,the maximum error in the reconstruction is 0.34mm.The accuracy of the free arc surface recognition by the rotating arc sensor meets the requirements. |
