摘要
对ARMA模型含有输入噪声和量测噪声的情况进行了讨论,在这两种情况下,最小二乘法并不能得到很好的估计结果,运用辅助变量法(IV)可以提高估计精度。特别是在输入信号受到干扰时,本文提出了加权辅助变量法(WIV),可以得到比IV方法更精确的结果,尤其是在噪信比(NSR)较高的场合。
In this paper, parameter estimation of ARMA system with input noise and measurement noise is respectively considered. The noise are supposed to be Gaussnoise with zero mean. In these cases, the instrumnetal variable method(IV) is used to get a good approach, whereas the LS method is not fit. Furthermore, for input noisy data the weighted instrumental variable (WIV) method is proposed to reduce the inaccuracy caused by high NSR. Finally, several simulation results are presented to show the advantage of the estimation algorithm discussed here.
出处
《南京大学学报(自然科学版)》
CAS
CSCD
1997年第1期25-31,共7页
Journal of Nanjing University(Natural Science)
基金
国家自然科学基金
关键词
噪声
辅助变量法
ARMA模型
系统辨识
参数估计
input noise
measurement noise
instrumental variable method
weighted instrumental variable method