统计检验抗差估计新方法研究  

Research of a new robust estimation based on statistical test

作　　者：郑勇波 周命端[2] 文鸿雁[3] 

机构地区：[1]上海市城市建设设计研究院,上海200125 [2]武汉大学测绘学院,武汉430079 [3]桂林理工大学土木工程系,广西桂林541004

出　　处：《测绘科学》2010年第5期72-73,共2页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(40574002)

摘　　要：20世纪60年代数学家重新从广泛的意义上提出非最小二乘估计问题,并已在抗差估计中取得重要的进展。从1980年起,抗差估计理论开始被引入测量领域,并出现了许多方案。本文依据统计检验,结合IGG1区间划分的部分思想,提出了一种全新的定权思路,给出了一个简要数学模型。利用MATLAB编程,将本文提出的方法与经典的IGG1方法进行了比较,对它的抗差效果进行了总结分析,通过一个经典算例,表明新方法具有准确定位粗差的优点。In the 1960s,mathematicians resumed the non-least-squares problems and have made prominent progress in the robust estimation.From 1980,robust estimation theory has been applied in the field of measuring,and there have been many schemes.According to statistical test and part of the division of thinking for IGG1 method,a new robust scheme with simple mathematical model was proposed in this paper.On a small net,the new scheme was compared with IGG1 method to find its unique merit of locating outlier.At last this paper made a summary of the results of trial computation.Via a classic example,it demonstrated that the new method had the advantage of accurately positioning the gross errors.

关 键 词：抗差估计 抗差 改正数 统计检验 粗差 

分 类 号：P207[天文地球—测绘科学与技术]