Detail publikace

Aircraft equipment modeling using neural networks

Originální název

Aircraft equipment modeling using neural networks

Anglický název

Aircraft equipment modeling using neural networks

Jazyk

en

Originální abstrakt

In this paper, the neural network module for simulating the behavior of an arbitrary system is described. The black-box modeling tool is aimed to simulate unknown systems characterized by measurements. The tool is able to model systems that cannot be described analytically due to their nonlinearity or complexity. Neural networks are suitable for solving real EMC problems leading to extremely high computation demands or to complicated experimental measurements. There are two main limitations: the sparsity of numerical results (due to the computation complexity) and the noise corrupting the measurement results. It is essential to find an equivalent model approximating the behavior of investigated system if the sparse or noisy data are available.

Anglický abstrakt

In this paper, the neural network module for simulating the behavior of an arbitrary system is described. The black-box modeling tool is aimed to simulate unknown systems characterized by measurements. The tool is able to model systems that cannot be described analytically due to their nonlinearity or complexity. Neural networks are suitable for solving real EMC problems leading to extremely high computation demands or to complicated experimental measurements. There are two main limitations: the sparsity of numerical results (due to the computation complexity) and the noise corrupting the measurement results. It is essential to find an equivalent model approximating the behavior of investigated system if the sparse or noisy data are available.

Dokumenty

BibTex


@inproceedings{BUT74514,
  author="Jitka {Svobodová} and Vlastimil {Koudelka} and Zbyněk {Raida}",
  title="Aircraft equipment modeling using neural networks",
  annote="In this paper, the neural network 
module for simulating the behavior of an 
arbitrary system is described. The black-box 
modeling tool is aimed to simulate unknown 
systems characterized by measurements. The tool 
is able to model systems that cannot be described 
analytically due to their nonlinearity or 
complexity. 
Neural networks are suitable for solving real 
EMC problems leading to extremely high 
computation demands or to complicated 
experimental measurements. There are two main 
limitations: the sparsity of numerical results (due 
to the computation complexity) and the noise 
corrupting the measurement results. It is essential 
to find an equivalent model approximating the 
behavior of investigated system if the sparse or 
noisy data are available.",
  address="COREP",
  booktitle="Proceedings of 2011 International Conference on Electromagnetics in Advanced Applications",
  chapter="74514",
  howpublished="electronic, physical medium",
  institution="COREP",
  year="2011",
  month="september",
  pages="627--630",
  publisher="COREP",
  type="conference paper"
}