DocumentCode :
815592
Title :
An additional requirement for innovations testing in system identification
Author :
Martin, W.C. ; Stubberud, A.R.
Author_Institution :
IBM Corporation, Thousand Oaks, CA, USA
Volume :
19
Issue :
5
fYear :
1974
fDate :
10/1/1974 12:00:00 AM
Firstpage :
583
Lastpage :
584
Abstract :
In [1], Boozer and McDaniel showed by means of a counterexample that whitening the measurement residual sequence output from the Kalman filter is not sufficient for identifying the transition matrix of a linear dynamic system. Recent work by the present authors, documented in [2], has shown that in many cases the identification can be accomplished by requiring further that the residual sequence have a zero mean. Imposing this requirement, the transition matrix of the Boozer and McDaniel example is identified.
Keywords :
Innovations methods; Kalman filtering; Linear systems, time-invariant discrete-time; System identification; Automatic control; Feedback control; Filters; Optimal control; Performance analysis; Satellites; Space vehicles; System identification; System testing; Technological innovation;
fLanguage :
English
Journal_Title :
Automatic Control, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9286
Type :
jour
DOI :
10.1109/TAC.1974.1100656
Filename :
1100656
Link To Document :
بازگشت