Robust system identification for non-persistently exciting input signals

A. W. Hull, W. K. Jenkins

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The performance of adaptive identification algorithms is constrained by the spectral characteristics of the input signals. Too few frequency components may result in parameter wander, and possible instability. Direct sequence spread-spectrum techniques may be used to increase the spectral richness of the training signal. Computer simulations of gradient descent algorithms for FIR (finite impulse response) and IIR (infinite impulse response) systems indicate that parameter wander is eliminated and the rate of convergence is dramatically increased. The convergence of least squares algorithms is unaffected, but it is conjectured that their numerical properties are improved.

Original languageEnglish (US)
Title of host publicationMidwest Symposium on Circuits and Systems
PublisherPubl by IEEE
Pages602-604
Number of pages3
StatePublished - 1990
EventProceedings of the 32nd Midwest Symposium on Circuits and Systems Part 2 (of 2) - Champaign, IL, USA
Duration: Aug 14 1989Aug 16 1989

Other

OtherProceedings of the 32nd Midwest Symposium on Circuits and Systems Part 2 (of 2)
CityChampaign, IL, USA
Period8/14/898/16/89

Fingerprint

Identification (control systems)
Impulse response
Computer simulation

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering
  • Electronic, Optical and Magnetic Materials

Cite this

Hull, A. W., & Jenkins, W. K. (1990). Robust system identification for non-persistently exciting input signals. In Midwest Symposium on Circuits and Systems (pp. 602-604). Publ by IEEE.
Hull, A. W. ; Jenkins, W. K. / Robust system identification for non-persistently exciting input signals. Midwest Symposium on Circuits and Systems. Publ by IEEE, 1990. pp. 602-604
@inproceedings{3a9a21c72c914f92b12d6b6631b5e158,
title = "Robust system identification for non-persistently exciting input signals",
abstract = "The performance of adaptive identification algorithms is constrained by the spectral characteristics of the input signals. Too few frequency components may result in parameter wander, and possible instability. Direct sequence spread-spectrum techniques may be used to increase the spectral richness of the training signal. Computer simulations of gradient descent algorithms for FIR (finite impulse response) and IIR (infinite impulse response) systems indicate that parameter wander is eliminated and the rate of convergence is dramatically increased. The convergence of least squares algorithms is unaffected, but it is conjectured that their numerical properties are improved.",
author = "Hull, {A. W.} and Jenkins, {W. K.}",
year = "1990",
language = "English (US)",
pages = "602--604",
booktitle = "Midwest Symposium on Circuits and Systems",
publisher = "Publ by IEEE",

}

Hull, AW & Jenkins, WK 1990, Robust system identification for non-persistently exciting input signals. in Midwest Symposium on Circuits and Systems. Publ by IEEE, pp. 602-604, Proceedings of the 32nd Midwest Symposium on Circuits and Systems Part 2 (of 2), Champaign, IL, USA, 8/14/89.

Robust system identification for non-persistently exciting input signals. / Hull, A. W.; Jenkins, W. K.

Midwest Symposium on Circuits and Systems. Publ by IEEE, 1990. p. 602-604.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Robust system identification for non-persistently exciting input signals

AU - Hull, A. W.

AU - Jenkins, W. K.

PY - 1990

Y1 - 1990

N2 - The performance of adaptive identification algorithms is constrained by the spectral characteristics of the input signals. Too few frequency components may result in parameter wander, and possible instability. Direct sequence spread-spectrum techniques may be used to increase the spectral richness of the training signal. Computer simulations of gradient descent algorithms for FIR (finite impulse response) and IIR (infinite impulse response) systems indicate that parameter wander is eliminated and the rate of convergence is dramatically increased. The convergence of least squares algorithms is unaffected, but it is conjectured that their numerical properties are improved.

AB - The performance of adaptive identification algorithms is constrained by the spectral characteristics of the input signals. Too few frequency components may result in parameter wander, and possible instability. Direct sequence spread-spectrum techniques may be used to increase the spectral richness of the training signal. Computer simulations of gradient descent algorithms for FIR (finite impulse response) and IIR (infinite impulse response) systems indicate that parameter wander is eliminated and the rate of convergence is dramatically increased. The convergence of least squares algorithms is unaffected, but it is conjectured that their numerical properties are improved.

UR - http://www.scopus.com/inward/record.url?scp=0025661814&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0025661814&partnerID=8YFLogxK

M3 - Conference contribution

SP - 602

EP - 604

BT - Midwest Symposium on Circuits and Systems

PB - Publ by IEEE

ER -

Hull AW, Jenkins WK. Robust system identification for non-persistently exciting input signals. In Midwest Symposium on Circuits and Systems. Publ by IEEE. 1990. p. 602-604