ROMANIAN JOURNAL OF INFORMATION SCIENCE AND
TECHNOLOGY
Volume 3, Number 1, 2000, 233 - 242
Modeling of the Membrane
Supported Microwave
Transmission Lines using
Electromagnetically Trained Artificial Neural Networks
Dan Neculoiu
"Politehnica" University of Bucharest, Romania
Sergiu Iordanescu, Mircea
Dragoman, Alexandru Muller
National Institute for R&D in Microtechnologies,
Bucharest, Romania
Abstract.
The paper presents electromagnetically (EM)
trained artificial neural networks (ANN) models for some membrane supported microwave
components as coplanar waveguides transmission lines, grounded coplanar waveguides and
microstrip lines. Full-wave electromagnetic simulation has been employed to characterize
the microwave components in terms of characteristic impedance and effective dielectric
permittivity. The ANN architecture used in this work consists of an input layer, an output
layer and one hidden layer. The hidden layer incorporates a nonlinear activation function
that allows modeling the complex input/output relationships. Advanced features for ANN
modeling where used to decrease the model development time. |