Research On Curve And Surface Reconstruction For Twisted Blade Based On Limited Measure Points

With the rapid development of energy, transportation, national defense and other fields, the demand for high precision machining of large, thin-walled and complex curved surface parts has become more and more urgent.As one of the typical representative, the processing level of pipe twisted blade will directly affect China’s aerospace and defense technology. Due to the twisted blade’s traditional way of processing procedure is complex and the rejection rate is high, the expensive titanium alloy hollow wide chord blade material of the twisted blade will lead to high processing costs. The development of adaptive processing system for the twisted blade can effectively solve this problem. Adaptive processing system is a system which integrates the functions of rough measurement, model reconstruction, process analysis, NC code generation and simulation. At present, the adaptive processing system in our country still needs further res earch and development.In this paper, in order to develop the model reconstruction function in adaptive processing system, the curve and surface reconstruction method of twist blade based on limited measurement points is studied to accurately reproduce the shape of the blade surface machining and to obtain more accurate simulation and processing results.Firstly, according to the requirements of adaptive processing system of twisted blade model reconstruction technology, the advantages and disadvantages of existing model reconstruction algorithms is analyzed.After that, the NURBS curve and surface model is chosen to reconstruct the curved surface of the blade. And the existing NURBS curve and surface fitting method is improved according to the specific reconstruction index.Secondly, before the reconstruction by NURBS curve and surface model, according to the necessary test data preprocessing step and the distributive characteristics of twisted blade measurement, sorting algorithm of measurement based on the quad tree partitioning is proposed, which realize the sorting of measurement point on section line efficiently and accurately. On the basis of measuring point arrangement, the slice method of measuring point data is given, which is the groundwork for the theoretical research of the following surface reconstruction.Thirdly, according to the curvature characteristics of twisted blade section line, the node vector parameterization method and the interpolation algorithm of NURBS curves and surfaces are discussed. In order to obtain the reconstructed model which can be as much as possible replicate the real processing parts, take the fitting error and curvature fitting error as the evaluation criterion of error. Based on the research of value data-point weight factor’s geometric meaning and its influence on the interpolation curve, a correction algorithm of the type of value data-point weight factor is proposed. This algorithm can reduce the error between the fitting surface and the actual parts surface.At last, based on above research results, the model reconstruction software is compiled based on the UG visualization platform. Related instances show that the reconstructed model can meet the requirements of the fitting error in the relevant parts on the basis of accurate reproduction type.