稳健再加权总体最小二乘法  

Robust re-weighting total least squares

作　　者：闫国庆 孙同贺[2] 燕志明[2] 

机构地区：[1]江门市勘测院有限公司,广东江门529000 [2]内蒙古科技大学矿业与煤炭学院,内蒙古包头014010

出　　处：《工程勘察》2016年第8期67-70,共4页Geotechnical Investigation & Surveying

摘　　要：当观测向量和系数矩阵均含有粗差时,本文提出将稳健WTLS和方差分量估计方法二者结合起来,利用方差分量技术合理地估计观测向量和系数矩阵的随机模型,重新定权,进而利用稳健WTLS方法进行参数解算,此方法迭代进行。亦用直线拟合实例证明了此方法的优越性。When observation vector and coefficient matrix contains gross errors,a combining method of robust WTLS and variance component estimation is proposed. Variance component estimation is used to choose stochastic models of observation vector and coefficient matrix, and robust weights are then computed. The solutions are obtained by the robust WTLS method. The superiority of the method is demonstrated by the line fitting.

关 键 词：稳健估计 稳健WTLS 方差分量估计 总体最小二乘 直线拟合 

分 类 号：O241.5[理学—计算数学]