摘要
针对总体最小二乘中粗差的可区分性,在Partial-EIV模型加权总体最小二乘算法的基础上引入了两个备选假设下的可靠性理论,给出了分析总体最小二乘粗差可区分性的方法。通过直线拟合的算例分析,说明本文的方法是可行的,能够有效地分析总体最小二乘中粗差的可区分性,并发现采用总体最小二乘求解直线拟合时,存在粗差不可区分的情况,也就意味着粗差是不可定位的。对于其它计算模型也可能存在粗差不可区分的情况,须加以注意。
In order to analysis distinguishability of gross error in total least square. In the paper the Partial-EIV model weighted total least square is applied. And the theory of reliability under two alternative hypotheses is cited. Thus, The method of analysising distinguishability of gross error in total least square is proposed. At last the example of linear fitting is carried out and illustrate that the method is feasibility. It is effective to analysis distinguishability of gross error in total least square. And the condition of indistin- guishable of gross error is existing when applying total least square to solve linear fitting. It＇s meaning that the gross error can not be located. It is need to be careful that the condition maybe exist in other calculation mode.
出处
《测绘科学》
CSCD
北大核心
2017年第7期46-51,共6页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41204003
41374007
41464001)
江西省科技落地计划项目(KJLD12077)
江西省教育厅科技项目(GJJ13457)
中国博士后基金项目(94773)
江西省中青年教师发展计划访问学者专项项目(2012-132)
江西省远航工程计划项目(2013-132)
关键词
总体最小二乘
相关系数
可区分性
粗差
total least square
correlation coefficient
distinguishability
gross error
