具有 LMS 稳健初值的选权迭代法

摘要

This paper explores the selecting weight iteration method with initial value by LMS ,which uses LMS robust estimate to asertain the initial residuals value .The selecting weight iteration (SWI) process goes after that .The estimate result of the method takes on high breakdown point robustness of LMS and high estimate effciency of SWI .The result shows the selecting weight iteration method with LMS estimation can resist outliers like the LS with no outliers .So it is more robust .%基于选权迭代法的基本理论，文中提出先用LMS稳健估计来确定残差的初值，然后再进行选权迭代方法。其估计结果既继承LMS方法的高失效点（BP）稳健性，又具有选权迭代法的高估计效率，其计算结果与无异常点时最小二乘估计结果基本一致。