岭型主成分估计与岭估计在抗差中的对比  

Comparison of ridge principal component estimation and ridge estimation in robust

作　　者：肖星星 王晓红 Xiao Xingxing;Wang Xiaohong(College of Mining EngineeringNorth China University of Science and Technology,Hebei Tangshan063210)

机构地区：[1]华北理工大学矿业工程学院,河北唐山062000

出　　处：《南方农机》2020年第23期51-53,共3页

摘　　要：岭估计主要用于减弱或消除数据呈病态性对参数估值的影响,但它依然存在缺陷,本文在岭估计的缺陷上,运用主成分估计方法,对平差Gauss-Markov参数模型进行改进,提出来一种新的有偏估计方法,称为岭型主成分组合估计,对岭型主成分组合估计、岭估计与最小二乘估计做了比较。结果表明数据呈严重病态时,岭型主成分组合估计和岭估计均方误差都小于LS估计,且岭型主成分组合估计的均方误差最小,表明岭型主成分组合估计和岭估计一样都可以改善LS估计,且其效果还优于岭估计。Ridge estimation is mainly used to weaken or eliminate the influence of ill conditioned data on parameter estimation.However,due to its defects,this paper uses principal component estimation method to improve the adjustment Gauss Markov parameter model,and proposes a new biased estimation method,called ridge principal component combination estimation The least square estimation is compared.The results show that when the normal equation is seriously ill conditioned,the mean square error of ridge principal component combination estimation and ridge estimation is less than LS estimation,and the mean square error of ridge principal component combination estimation is the smallest.It shows that both ridge principal component combination estimation and ridge estimation can improve LS estimation,and its effect is better than ridge estimation.

关 键 词：岭估计 Gauss-Markov模型 岭型主成分组合估计 最小二乘估计 

分 类 号：P420.4[天文地球—大气科学及气象学] P207