Publication detail

Evaluation of Electromagnetic Immunity of Layered Structures by Neural Networks

KOUDELKA, V. RAIDA, Z.

Original Title

Evaluation of Electromagnetic Immunity of Layered Structures by Neural Networks

English Title

Evaluation of Electromagnetic Immunity of Layered Structures by Neural Networks

Type

journal article - other

Language

en

Original Abstract

In this paper, exploitation of artificial neural networks for an efficient solution of a simple electromagnetic compatibility problem is discussed. Two parallel dielectric layers are penetrated by the perpendicular electromagnetic wave. A standing wave is formed between layers. Radial basis function networks are employed to estimate the electric field intensity between the layers for both of the harmonic wave and pulse wave illumination. The electrical parameters of dielectric layers can influence the field distribution inside the investigated structure. Probabilistic neural networks are used to classify parameters of layers related to critical intensities of internal fields. Classification abilities of probabilistic networks are compared with a conventional k-NN method both for a dense training set and a sparse one.

English abstract

In this paper, exploitation of artificial neural networks for an efficient solution of a simple electromagnetic compatibility problem is discussed. Two parallel dielectric layers are penetrated by the perpendicular electromagnetic wave. A standing wave is formed between layers. Radial basis function networks are employed to estimate the electric field intensity between the layers for both of the harmonic wave and pulse wave illumination. The electrical parameters of dielectric layers can influence the field distribution inside the investigated structure. Probabilistic neural networks are used to classify parameters of layers related to critical intensities of internal fields. Classification abilities of probabilistic networks are compared with a conventional k-NN method both for a dense training set and a sparse one.

Keywords

dielectric walls, electromagnetic modeling, feed forward neural network, radial basis functions, probabilistic neural network

RIV year

2011

Released

21.03.2011

Pages from

482

Pages to

489

Pages count

7

BibTex


@article{BUT50007,
  author="Vlastimil {Koudelka} and Zbyněk {Raida}",
  title="Evaluation of Electromagnetic Immunity of Layered Structures by Neural Networks",
  annote="In this paper, exploitation of artificial neural networks for an efficient solution of a simple electromagnetic compatibility problem is discussed. Two parallel dielectric layers are penetrated by the perpendicular electromagnetic wave. A standing wave is formed between layers. Radial basis function networks are employed to estimate the electric field intensity between the layers for both of the harmonic wave and pulse wave illumination. The electrical parameters of dielectric layers can influence the field distribution inside the investigated structure. Probabilistic neural networks are used to classify parameters of layers related to critical intensities of internal fields. Classification abilities of probabilistic networks are compared with a conventional k-NN method both for a dense training set and a sparse one.",
  chapter="50007",
  number="4",
  volume="vol. 5",
  year="2011",
  month="march",
  pages="482--489",
  type="journal article - other"
}