Detail publikace

A Reverse Neural Model of a General Planar Transmission Line

Originální název

A Reverse Neural Model of a General Planar Transmission Line

Anglický název

A Reverse Neural Model of a General Planar Transmission Line

Jazyk

en

Originální abstrakt

In the paper, an original exploration of neural networks for the design of planar microwave transmission lines is discussed. Since no analytical models of these transmission lines are at the disposal, an accurate finite-element analysis of a certain number of selected structures is performed and obtained results are used as learning patterns for training a neural network. In order to obtain a design tool, inputs of the network are associated with technical parameters of the transmission line, and outputs correspond to its physical parameters. Training of the network is based on a novel genetic algorithm, which ensures low error and good convergence of the learning process. Validity of the model is verified using previously published data.

Anglický abstrakt

In the paper, an original exploration of neural networks for the design of planar microwave transmission lines is discussed. Since no analytical models of these transmission lines are at the disposal, an accurate finite-element analysis of a certain number of selected structures is performed and obtained results are used as learning patterns for training a neural network. In order to obtain a design tool, inputs of the network are associated with technical parameters of the transmission line, and outputs correspond to its physical parameters. Training of the network is based on a novel genetic algorithm, which ensures low error and good convergence of the learning process. Validity of the model is verified using previously published data.

Dokumenty

BibTex


@inbook{BUT54990,
  author="Zbyněk {Raida}",
  title="A Reverse Neural Model of a General Planar Transmission Line",
  annote="In the paper, an original exploration of neural networks for the design of planar microwave transmission lines is discussed. Since no analytical models of these transmission lines are at the disposal, an accurate finite-element analysis of a certain number of selected structures is performed and obtained results are used as learning patterns for training a neural network. In order to obtain a design tool, inputs of the network are associated with technical parameters of the transmission line, and outputs correspond to its physical parameters. Training of the network is based on a novel genetic algorithm, which ensures low error and good convergence of the learning process. Validity of the model is verified using previously published data.",
  address="Physica-Verlag (A Springer-Verlag Company)",
  booktitle="The State of the Art in Computational Intelligence",
  chapter="54990",
  edition="Advances in Soft Computing",
  institution="Physica-Verlag (A Springer-Verlag Company)",
  year="2000",
  month="august",
  pages="203",
  publisher="Physica-Verlag (A Springer-Verlag Company)",
  type="book chapter"
}