Detail publikace

Multi-Dimensional Predictive Control of Hot-Air Tunnel Using a Neural Network Modell

PIVOŇKA, P., NEPEVNÝ, P.

Originální název

Multi-Dimensional Predictive Control of Hot-Air Tunnel Using a Neural Network Modell

Typ

článek ve sborníku ve WoS nebo Scopus

Jazyk

angličtina

Originální abstrakt

Model-based predictive controller (MPC) is a kind of optimal controller based on system model. Model is used for prediction of future system output and it is used for finding an optimal control action. Control action is always optimal according to the given criterion and constraints, which can be directly implemented into control algorithm. A feed-forward neural network with back-propagation learning algorithm is used as a system model. Obtained controller is adaptive, because the neural network model is able to observe system changes and adapt itself. This paper presents using of a multi-dimensional model-based predictive controller for hot air tunnel control. Two quantities of hot-air tunnel are controlled - the air flow and the temperature. These quantities are controlled in feedback control loops. The algorithm was implemented in MATLAB-Simulink and tested on the physical model. Communication between PC and hot-air tunnel is provided by PLC (connected via Ethernet). The practical results are discussed and advantages and disadvantages of multi-dimensional model-based predictive controller are shown.

Klíčová slova

Model-based predictive controller, Neural network model, PLC

Autoři

PIVOŇKA, P., NEPEVNÝ, P.

Rok RIV

2005

Vydáno

19. 9. 2005

Nakladatel

R. Hampel

Místo

Zittau

ISBN

3-9808089-6-3

Kniha

12th Zittau East-West Fuzzy Colloquium

Strany od

176

Strany do

181

Strany počet

6

BibTex

@inproceedings{BUT15252,
  author="Petr {Pivoňka} and Petr {Nepevný}",
  title="Multi-Dimensional Predictive Control of Hot-Air Tunnel Using a Neural Network Modell",
  booktitle="12th Zittau East-West Fuzzy Colloquium",
  year="2005",
  pages="6",
  publisher="R. Hampel",
  address="Zittau",
  isbn="3-9808089-6-3"
}