Localization of buried object using Backpropagation Nueral Network

This paper presents a BackPropagation Neural Network (BPNN) approach to predict the location of a buried object. A small circuit board is buried in sand at three different location levels and an electrically-small probe is used as a detection sensor. The standard deviation of the phase of reflection coefficient is used as input for the Neural Network (NN) while the location levels of the small circuit are designated as the output of the network. The network shows an accuracy of more than 90% in predicting the location of the buried circuit.