Research On Measured Data Processing And Error Evaluation Method Of Twisted Free-form Surface

The complex free-form surfaces, blisk and blade of aero engine included, arebig-twisted, they have more manufacture difficulties than the common free-formsurface, but the complex part decides the overall performance of the equipment, eveneffectes the technological level and national defense capabilities. Existing methods offree-form surface measurement、surface fitting and error evaluation mainly targetedthe common free-form surface, some methods in business software and literature wereineffective for twisted free-form surface, and barely studied for the surface withbig-twist and thin edge. So the study on data processing and evaluation for twistedfree-form surface is very important for improving the accuracy of complex part andtechnological level of our country.For the twisted free-form surface, this paper studied the data preprocessing, thecontinued montage of free-form surface, fitting and error evaluation under theuniform coordinate system, based on the point cloud measured by the on-line in-situhigh-precision machine. The main work is as follows:The structural design and error calibration of on-line in-situ measuring machinedeveloped by State Key Laboratory of precision measurement technology andinstruments in Tianjin University was described, and the error compensation ofthermal deformation and optical distortion was proposed the measuring machinecould measure all kinds of complex twisted free-form surfaces, including the surfacewas smaller than1000mm in diameter, with the gap between patches only10mm, thetorsion angle can be65°,and the measuring error was no more than0.01mm.The deleting method of abnormal noises is researched. Ineorporating the3δ lawof mathematical statistics, by means of calculating mean and standard error todeterminate threshold, the adaptive counting threshold method is putforward, whichhas well adaptivity. Denoising cloudy data based wavelet technology is researched.Using wavelet technology to denoise is the process of decomposing signal withwavelet, acquiring approximate coeffieients and detail coeffieients, acting detailedcoefficients with soft threshold to suppress noises, then reconstructing signal to getthe goal of denoising. The key of simplifying data technique is retaining originalfeatures of cloudy data utmost. On the basis of reseaching current simplification technique, new method using second generation wavelet based on lifting tosimplification data is proposed.TheG1smooth merging theory of the piecewise Bezier surface is analyzed, theBezier curve is a special NURBS curve is considered, the idea of resort to BeziersurfaceG1continuous condition for solving theG1continuous condition ofNURBS surfaces is presented.G1continuous condition of the NURBS is deducedand realized. Linear system ofG1continuous condition is solved, which providedthe basis for the following surface fitting.The existing grid generation theory is studied, incapability of grid generation thatthese theory to the four sided region constructed by four discontinuous broken line isfound. A rapid parametrization algorithm which is used to spatial four-sided region ontriangular mesh is presented in the paper. By comparing two kinds of parametrizationprojects of the spatial four-sided region, directly grid generation algorithm is chose. Aplanar two-directional flexing grid generation algorithm is presented which canquickly parametrizing the planar four-sided region, in fact, a spatial four-sided regionis parametrizated. Comparing two NURBS surface fitting methods, in order toenhance the fitting precision and obtain more detail information of the NURBSsurface fitting, two-directional fitting method is adopted to fitting the NURBS surface.For the points on the common boundaries and corners is restricted base on theG1smooth connection condition. So by fitting every patches and a wholeG1smoothsurface is obtained.Based on non-uniform B-splines, a new algorithm of calculating the normalvector at random position was deduced, and a formula with probe radius correctionwas proposed. An optimization algorithm named simplex method was applied in orderto eliminating the orientation error in the stage of workpiece measurement. It cansettle surfaces matching well by shifting and rotating the measuring coordinatesystem.