摘要
抗差估计方法是一种很好的抵抗粗差的方法。但是,不同的抗差估计方法具有不同的抗差特性。文章以三种不同网形、不同观测值数量和粗差数量、不同粗差数值的水准网为例,采用仿真实验的方法比较了观测值不等权或者服从广义高斯分布情况下常用13种抗差估计方法的抗差特性。研究结果表明,抗差特性更优的估计方法为L1法、Danish法、Geman-McClure法和IGGIII方案。
Robust estimation method is commonly used method to eliminate or weaken gross errors in surveying data processing.However,different robust estimation methods may have different robustness.Taking 3 kinds of leveling networks with different number of observations and gross errors as an example,the simulation experiments have been employed to compare the robustness of 13 robust estimation methods under the condition of observations having unequal weights or obeying generalized Gauss distribution.The results of simulation study indicate that L1 method,Danish method,Geman-McClure method and IGGIII scheme have better robustness and higher estimation efficiency.
出处
《北京测绘》
2017年第5期6-10,共5页
Beijing Surveying and Mapping
基金
江西省教育厅科技研究项目(GJJ151489)
山西省自然科学基金项目(2012011015-2)
关键词
水准网
广义高斯分布
不等权
抗差估计
方法比较
leveling network
generalized Gauss distribution
unequal weights
robust estimation
comparison of methods
