Detail publikace

Applications of Neural Networks in Real-Time Identification

Originální název

Applications of Neural Networks in Real-Time Identification

Anglický název

Applications of Neural Networks in Real-Time Identification

Jazyk

en

Originální abstrakt

Identification of dynamic systems is essential for adaptive control. Obviously, we have to obtain information on the dynamic behavior of the whole system - to identify it. One approach is the monitoring of the system characteristics, refinement and thus eliminating potential changes. The most frequently used method is the least squares method. Its advantage is the fast determination of the sought-for parameters, however, we are limited by the choice of the suitable sampling period T0. Making sampling period even shorter leads to unrealistic estimates of the state. The neural network seems to be a desirable solution. The neural network application in progress identification, mainly the Levenberg-Marquardt training algorithm is discussed.

Anglický abstrakt

Identification of dynamic systems is essential for adaptive control. Obviously, we have to obtain information on the dynamic behavior of the whole system - to identify it. One approach is the monitoring of the system characteristics, refinement and thus eliminating potential changes. The most frequently used method is the least squares method. Its advantage is the fast determination of the sought-for parameters, however, we are limited by the choice of the suitable sampling period T0. Making sampling period even shorter leads to unrealistic estimates of the state. The neural network seems to be a desirable solution. The neural network application in progress identification, mainly the Levenberg-Marquardt training algorithm is discussed.

BibTex


@inproceedings{BUT11418,
  author="Jiří {Dohnal} and Petr {Pivoňka}",
  title="Applications of Neural Networks in Real-Time Identification",
  annote="Identification of dynamic systems is essential for adaptive control. Obviously, we have to obtain information on the dynamic behavior of the whole system - to identify it. One approach is the monitoring of the system characteristics, refinement and thus eliminating potential changes. The most frequently used method is the least squares method. Its advantage is the fast determination of the sought-for parameters, however, we are limited by the choice of the suitable sampling period T0.  Making sampling period  even shorter leads to unrealistic estimates of the state. The neural network seems to be a desirable solution. The neural network application in progress identification, mainly the Levenberg-Marquardt training algorithm is discussed.",
  address="Rektor der Hochschule Zittau/Görlitz",
  booktitle="Proceedings East West Fuzzy Colloquium 2004",
  chapter="11418",
  institution="Rektor der Hochschule Zittau/Görlitz",
  year="2004",
  month="september",
  pages="156",
  publisher="Rektor der Hochschule Zittau/Görlitz",
  type="conference paper"
}