作　　者： ; ; ; ;

机构地区： 北京航空航天大学交通科学与工程学院

出　　处： 《机械设计》 2010年第2期7-10,共4页

摘　　要： 提出改进的移动最小二乘(MLS)截面数据拟合算法,改善传统MLS拟合方法导致尖锐特征消失情况。首先引入与切矢夹角相关的权函数来改善原来的距离权函数。取计算点的切矢向量为其最近节点的切矢向量,当节点切矢向量与计算点切矢向量夹角大于阈值,该节点角度权值为0,否则为1。最终的权函数为距离权函数与角度权函数的乘积。其次,通过引入无穷小量,将特征点分解为两个无限与其靠近的数据点。经实验验证,当影响域半径在适当范围内变化时,改进的MLS方法均能很好地保持特征。 The fitting algorithm of improved moving least square （MSL） sectional date has been put forward so as to improve disappearing circs of sharp characteristics caused by the traditional MSL fitting method. Firstly introduced the weight function interrelated to the tangent vector included angle to improve the original distance weight function. Take the tangent vector of calculation point as tangent vector of its nearest node; while the included angle between tangent vector of node and tangent vector of calculation point was greater than the threshold value, the weighted value of this node angle was taken as O, otherwise as 1. The final weight function was the product of weighted function of distance and weighted function of angle. Secondly, by means of introducing the infinitesimal let the characteristic point be divided into two data points infinitely close to the characteristic point. Through experimental verification that while the radius of influencing domain varies within proper ranges the improved MLS method could all nicely maintain the characteristic.

关 键 词： 截面数据拟合 移动最小二乘 特征保持 角度权函数

领　　域： [自动化与计算机技术] [自动化与计算机技术]