Application of UWB Electromagnetic Waves for Subsurface Object Location Classification by Artificial Neural Networks

Abstract

The problem of determination of object position in a plane is solved by the analysis of ultrawideband electromagnetic wave reflected from the subsurface object. The model of ground containing perfectly conducting object inside is irradiated by short impulse wave with Gaussian time dependence. The direct problem is solved by FDTD method to receive a time dependence of reflected wave amplitude. To recognize the presence of the object and depths of its position the multilayer artificial neural networks (ANN) is used. The amplitudes of electric component of the reflected field in different time and special points above the ground surface are the input data for multilayer ANN of different structures. The work of the trained ANN is verified for arbitrary depths of object position.