Skip to content

Publications

Detail of publication

Citation

Šimandl, M. and Král, L. : Neural adaptive dual controller with dynamic structure . Controlo 2006, p. 1-6, Instituto Superior Tecnico, Lisbon, 2006.

Abstract

A new neural adaptive dual controller with dynamic structure for nonlinear stochastic systems is designed. The model of the stochastic system is based on a Gaussian radial basis function neural network. The number of unknown parameters of the network grows rapidly with system dimension. In order to avoid this problem, an on-line dynamic structure algorithm of the network is proposed. The algorithm is founded on a suitable growing and pruning criterion. The dual controller design is based on two separate criteria, each introducing one of opposing aspects of estimation and control; caution and probing. The proposed approach is illustrated in a numerical example.

Detail of publication

Title: Neural adaptive dual controller with dynamic structure
Author: Šimandl, M. ; Král, L.
Language: English
Date of publication: 11 Sep 2006
Year: 2006
Type of publication: Papers in proceedings of reviewed conferences
Title of journal or book: Controlo 2006
Page: 1 - 6
Publisher: Instituto Superior Tecnico
Address: Lisbon
Date: 11 Sep 2006 - 13 Sep 2006
/ /

Keywords

Neural networks, adaptive control, stochastic control, nonlinear systems, parameter estimation, non-Gaussian processes, learning algorithms

BibTeX

@INPROCEEDINGS{SimandlM_2006_Neuraladaptivedual,
 author = {\v{S}imandl, M. and Kr\'{a}l, L.},
 title = {Neural adaptive dual controller with dynamic structure},
 year = {2006},
 publisher = {Instituto Superior Tecnico},
 journal = {Controlo 2006},
 address = {Lisbon},
 pages = {1-6},
 url = {http://www.kky.zcu.cz/en/publications/SimandlM_2006_Neuraladaptivedual},
}