摘要
针对抗差估计选权迭代法残差初值缺陷的求解问题,该文基于选权迭代法的基本理论深入分析了最小二乘估计、最小一范数估计和最小截平方和(LTS)3种残差初值求解方法,探究以3种方法求解残差初值为基础的选权迭代法的适用范围和抗差效果。通过实例验证表明,LTS方法残差初值具有较强的稳健性,基于LTS的选权迭代法抗差效果优于其他2种方法,具有较强的粗差探测能力。
For the problem of the initial residual value with the selecting weight iteration method, the paper analyzed the least square estimation, least absolute deviation and least trimmed squares （LTS）deep- ly based on the basic theory of selecting power iteration method, then explored three methods of selecting weight iteration to resist multiple outliers. The result showed that the initial residual value of LTS was more robust and the selecting weight iteration method with LTS estimation could resist outliers better than the others and detected the gross error with the strong ability.
出处
《测绘科学》
CSCD
北大核心
2015年第8期18-21,共4页
Science of Surveying and Mapping
基金
国家自然科学基金资助项目(41374007)
测绘地理信息江西省研究生创新教育基地项目
