Neural Network Application for Dielectric Structure Parameter Determination by Multifrequency Methods

Author

Drobakhin, O.O. ; Doronin, A.V.

Author_Institution

Dnepropetrovsk Nat. Univ.

fYear

2006

fDate

Sept. 2006

Firstpage

358

Lastpage

360

Abstract

Application of neural network ideology for layered dielectric structure nondestructive testing is under consideration. The 3-layered feed-forward with activation function as a sigmoid (S-shaped) was chosen. Time-domain signals synthesized from ultrawideband data is used as input signal. Levenberg-Marquardt algorithm was used for training of the network. Some results of numerical simulation for two-layered structure with air insertion in presence of Gaussian additive noise are demonstrated

Keywords

Gaussian noise; dielectric materials; feedforward neural nets; learning (artificial intelligence); multilayers; nondestructive testing; parameter estimation; signal synthesis; thickness measurement; ultra wideband technology; Gaussian additive noise; Levenberg-Marquardt algorithm; S-shaped activation function; air insertion; dielectric structure; multifrequency methods; multilayered structure; network training; neural network application; nondestructive testing; parameter determination; sigmoid activation function; thickness estimation; three-layered feed-forward network; time-domain signal synthesis; two-layered structure; ultrawideband signal; videopulse synthesis; Additive noise; Dielectrics; Feedforward systems; Network synthesis; Neural networks; Nondestructive testing; Numerical simulation; Signal synthesis; Time domain analysis; Ultra wideband technology; Multilayered structure; neural network; thickness estimation; time-domain synthesized videopulse; ultrawideband signal;

fLanguage

English

Publisher

ieee

Conference_Titel

Ultrawideband and Ultrashort Impulse Signals, The Third International Conference

Conference_Location

Sevastopol

Print_ISBN

1-4244-0513-0

Type

conf

DOI

10.1109/UWBUS.2006.307256

Filename

4099314