A weighted total least squares estimator for multivariable systems with nearly maximum likelihood properties

被引：5
|
作者：
Guillaume, P [1 ]
Pintelon, R
Schoukens, J
机构：
[1] Free Univ Brussels, Dept Mech Engn, B-1050 Brussels, Belgium
[2] Free Univ Brussels, Elect Measurement Dept, B-1050 Brussels, Belgium
来源：
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT | 1998年 / 47卷 / 04期
关键词：
errors-in-variables; (generalized) singular value decomposition; (generalized) total least squares; maximum likelihood; multivariable systems; parameter estimation; system identification;
D O I：
10.1109/19.744627
中图分类号：
TM [电工技术]; TN [电子技术、通信技术];
学科分类号：
0808 ; 0809 ;
摘要：
The aim of the present paper is to develop a parametric estimator for linear time-invariant multivariable systems with nearly maximum likelihood properties. The estimator is based on the total least squares (TLS) method. It can be seen as an "optimally" weighted iterative generalized TLS (GTLS) estimator, combining the nice asymptotic properties of the maximum likelihood (ML) method with the global minimization property of the GTLS estimator.
引用
收藏
页码：818 / 822
页数：5
相关论文
共 50 条
12345