一次范数最小和选权迭代联合的抗差法

A Robust Estimation Method Combined First Order Minimum Norm and Selecting Weight Iteration Method

联合一次范数最小和选权迭代法进行抗差估计。使用一次范数进行粗差定位,在确定粗差位置后,对相应的观测值进行降权,以此作为选权迭代法的初始权进行迭代平差,解决初始权不合理导致选权迭代法失效的问题。最后以水准网数据验证了该联合法的正确性。

If there are no ideas of the locations of gross errors,selecting weight iteration method runs the first adjustment with equal weight.It leads to gross errors transferred to other observations.The robust estimation will fail because of incorrect weight.The first order minimum norm has the ability to resist gross errors,but is not optimal under Gaussian distribution.Combined two methods above,which the first order minimum norm is used to determine the location of gross errors,and then selecting weight iteration method reduces weight of those gross errors,we propose a novel robust estimation.The novel method is validated by a level net data processing.