稳健估计权函数的对比研究

Comparison of Robust Estimation Weight Functions

稳健估计的思想相比于经典的最小二乘法更加适用于消除或减弱观测值中的粗差对变形监测数据计算的影响,然而不同稳健估计方法的稳健性不尽相同。以唐山市百货大楼固定水准监测数据为例,通过仿真实验对比常用稳健估计方法的稳健性。实验结果表明,对于独立等权观测值的水准观察数据,Huber权函数的方案较其它权函数能够更好地消除粗差。

Robust estimation is more suitable than the classical least squares method to eliminate or reduce the influence of the gross error in the observation on the calculation of deformation monitoring data. However,the robustness of different robust estimation methods is different.The fixed level monitoring data of Tangshan Department Store was taken as an example,and the robustness of the commonly used robust estimation methods was compared by simulation experiments.The experimental results show that the scheme of the Huber weight function can eliminate the gross error better than other weight functions for the level observation data of independent equal weight observations.