High Precision Reconstruction Of Curved Surface With Rapidly Changing Curvature Based On Point Cloud Data

With the rapid development of China in the fields of transportation,energy,national defense and so on,higher machining accuracy is required for complex parts with rapidly changing curved surfaces.In order to ensure the extremely high machining accuracy of this kind of parts,the measurement-machining integrated closed-loop control strategy is often adopted to reconstruct the high-precision curved surface of the measuring points and calculate the machining allowance to serve the tool path planning in the process of feedback compensation.At present,the surface reconstruction technology for curvature rapidly changing surface point cloud still has the following shortcomings: first,the surface shape of curved surface with rapidly changing curvature is complex,and the amount of measured data is huge,so it is easy to produce numerical fluctuation phenomenon when using high-order surface to reconstruct.Second,the curvature rapidly changing surface has the characteristic of uneven distribution of geometric features.It is easy to produce excess error in the area with complex surface geometry when using the conventional surface reconstruction method,which reduces the quality of surface reconstruction.In order to solve the above problems,this paper studies the NURBS surface reconstruction method for curvature rapidly varying surface scattered point cloud data,and puts forward the knot vector adaptive optimization algorithm and NURBS surface local optimization algorithm to realize the high-precision surface reconstruction of curvature rapidly varying surface point cloud.The specific research contents are as follows:(1)The knot vector adaptive optimization algorithm considering the uneven distribution of geometric features of complex surfaces is used to improve the approximation ability of reconstructed surfaces to complex geometric features.By constructing the geometric feature evaluation function to evaluate the complexity of the surface,the local surface complexity corresponding to each knot interval on the initial reconstructed surface is evaluated,and then the knots are inserted in the regions with complex geometric features.the particle swarm optimization algorithm is used to optimize the specific position of the inserted knots.Finally,the adaptive optimization of node vector considering geometric features is realized.(2)The local optimization algorithm of NURBS surface based on local control point adjustment.Firstly,the knot insertion technique is used to improve the flexibility of shape control of the regions with excessive errors;then,by using the local support properties of the NURBS surface,the local control points corresponding to such regins are extracted,and the multi-objective optimization function is found and solved by linear weighting method and quasi-Newton method.Finally,the local optimization of the super-difference region of NURBS surface is realized.(3)The graphical user interface is designed based on MATLAB-GUI,and the auxiliary software for surface reconstruction is formed.The operation interface of the reconstruction assistant software is made by using MATLAB-GUI,and the function files of knot vector adaptive optimization algorithm and NURBS surface local optimization algorithm are compiled respectively.The surface reconstruction software written in this paper integrates the functions of node vector adaptive optimization and local optimization.In this paper,the knot vector generated by the knot vector adaptive optimization algorithm considering the uneven distribution of geometric features of complex surfaces has dense nodes in the areas with complex geometric features.Combined with NURBS surface local optimization algorithm,it can significantly improve the reconstruction accuracy of reconstructed surfaces and realize high-precision surface reconstruction of point clouds of rapidly changing surfaces,which is of great significance to improve the manufacturing accuracy of curvature rapidly changing surface parts.