从混合正态分布的角度研究相关稳健估计函数模型  

The Study of Correlated Robust Estimation Function Model Based on Mixed Normal Distribution

作　　者：姚宜斌[1] 

机构地区：[1]武汉大学测绘科学与技术学院,湖北武汉430079

出　　处：《测绘工程》2001年第4期30-32,44,共4页Engineering of Surveying and Mapping

基　　金：国家自然科学基金资助项目(40004001)

摘　　要：传统的稳健估计方法，其核心在于利用直接定义■函数法来构造合理的稳健估计函数模型口 文中把含异常观测值的观测量看作一种混合正态分布，并按极大似然估计的定义式来定义相应的稳健估计模型。这种方法简单直观，易于理解，且所定义的稳健估计模型与直接定义■函数法所构造的基于等价方差一协方差的稳健最小二乘估计函数模型［１］完全一致。The core of traditional robust estimation method is the direct definition of■ function to es- tablish the robust estimation function model. In this paper, the observations which include some abnor- mal values were looked as mixed normal distribution, and the robust estimation numerical model also be given by the definition of M-estimation. This method is simple and easy to understand, and the robust estimation model it defined is consistent with the correlated robust estimation numerical model based on equivalent variance-covariance.

关 键 词：稳健估计 Ψ函数 混合正态分布 异常观测值 方差 误差 

分 类 号：P207[天文地球—测绘科学与技术]