Affiliations: [a] Laboratoire de Génie Electrique de Paris,
Université Paris-Sud, Université Pierre et Marie Curie, Gif-sur Yvette
cedex, France
Correspondence:
[*]
Corresponding author: M.K. Smail, Laboratoire de Génie
Electrique de Paris, UMR 8507 CNRS, SUPELEC, Université Paris-Sud,
Université Pierre et Marie Curie, 11 rue Joliot-Curie, 91192 Gif-sur Yvette
cedex, France. E-mail: mostafa. smail@lgep.supelec.fr
Abstract: A fast diagnosis dedicated to embedded transmission lines and based
on time domain reflectometry is presented. The forward problem allows to
simulate the propagation along the wiring network as well as to create datasets
for the inverse problem resolution. Neural networks (NNs) are used to solve the
inverse problem which consists to find defects on the wire from the
reflectometry response. The significant parameters of the reflectometry
response data are extracted using principal component analysis. This method
allows an efficient reduction of the dimension of the reflectometry data
space.
Keywords: Wiring network, multiconductor transmission lines, FDTD method, time domain reflectometry, reconstruction, neural network and principal component analysis
