Skip to content

Publications

Detail of publication

Citation

Král, L. and Šimandl, M. and Duník, J. and Hering, P. : Kalman filtering based methods for identification of nonlinear systems by neural networks . Proceedings of 6th International carpathian control conference, p. 345-350, University of Miskolc, Miskolc-Lillafüred, 2005.

Abstract

Identification of nonlinear system using neural networks and Kalman filtering based methods are treated. The stress is laid on networks parameters estimation, particularly by derivative-free Kalman filter based methods. The methods are easier for implementation because the Jacobian needed in standard methods as Extended Kalman Filters are not necessary to be evaluated. The derivative-free filters are compared with Extended Kalman filter, regarding especially estimation quality and numerical demands.

Detail of publication

Title: Kalman filtering based methods for identification of nonlinear systems by neural networks
Author: Král, L. ; Šimandl, M. ; Duník, J. ; Hering, P.
Language: English
Date of publication: 24 May 2005
Year: 2005
Type of publication: Papers in proceedings of reviewed conferences
Title of journal or book: Proceedings of 6th International carpathian control conference
Page: 345 - 350
ISBN: 963 661 643 4
Publisher: University of Miskolc
Address: Miskolc-Lillafüred
Date: 24 May 2005 - 27 May 2005
/ /

Keywords

System identification, nonlinear system, neural network training, multilayer perceptron network, derivative-free Kalman filter

BibTeX

@INPROCEEDINGS{KralL_2005_Kalmanfiltering,
 author = {Kr\'{a}l, L. and \v{S}imandl, M. and Dun\'{i}k, J. and Hering, P.},
 title = {Kalman filtering based methods for identification of nonlinear systems by neural networks},
 year = {2005},
 publisher = {University of Miskolc},
 journal = {Proceedings of 6th International carpathian control conference},
 address = {Miskolc-Lillaf\"{u}red},
 pages = {345-350},
 ISBN = {963 661 643 4},
 url = {http://www.kky.zcu.cz/en/publications/KralL_2005_Kalmanfiltering},
}