Abstract
A new robust and computationally efficient solution to least-squares problem in the presence of round-off errors is proposed. The properties of a harmonic regressor are utilized for design of new combined algorithms of direct calculation of the parameter vector. In addition, an explicit transient bound for estimation error is derived for classical recursive least-squares algorithm using the Lyapunov function method. Different initialization techniques of the gain matrix are proposed as an extension of the recursive least-squares algorithm. All the results are illustrated by simulations.
Keywords
Get full access to this article
View all access options for this article.
References
1.
Ljung
L.
System identification: theory for the user . Upper Saddle River, NJ : Prentice Hall , 1999 .
2.
Ljung
S
Ljung
L.
Error propagation properties of recursive least-squares adaptation algorithms . Automatica 1985 ; 21 (2 ): 157 –167 .
3.
Verhaegen
M
Van Dooren
P.
Numerical aspects of different kalman filter implementations . IEEE T Automat Contr 1986 ; 31 (10 ): 907 –917 .
4.
Liavas
A
Regalia
P.
On the numerical stability and accuracy of the conventional recursive least squares algorithm . IEEE T Signal Proces 1999 ; 47 (1 ): 88 –96 .
5.
Bayard
D.
A general theory of linear time-invariant adaptive feedforward systems with harmonic regressors . IEEE T Automat Contr 2000 ; 45 (11 ): 1983 –1996 .
6.
Stotsky
A.
Recursive trigonometric interpolation algorithms . Proc IMechE, Part I: J Systems and Control Engineering 2010 ; 224 (1 ): 65 –77 .
7.
Stotsky
A.
A new frequency domain system identification method . Proc IMechE, Part I: J Systems and Control Engineering 2012 ; 226 : 111 –124 .
8.
Avedyan
E.
Modified Kaczmarz algorithms for estimating the parameters of linear plants . Automat Rem Contr 1978 ; 39 (5 ): 674 –680 .
9.
Horn
R
Johnson
C.
Matrix analysis . Cambridge : Cambridge University Press , 1985 .
10.
Atkinson
K.
An introduction to numerical analysis . 2nd ed. New York : John Wiley & Sons , 1989 .
11.
Ye
Q.
Computing singular values of diagonally dominant matrices to high relative accuracy . Math Comput 2008 ; 77 (264 ): 2195 –2230 .