Dielectric Object Subsurface Survey by Ultrawideband Radar and ANN

The problem of subsurface object classification by ground penetrating radar with artificial neural network data processing is solving in the work. Antenna system generates ultrashort impulse electromagnetic field and receive the reflections from the investigated area of ground. Obtained signals are exempted from surface influence and normalized on the root of its energy for amplifying the useful part of the impulse. Neural network implements the detection of the objects in the angular sector relatively to antenna system. Asymmetrical dielectric mine response on impulse electromagnetic wave is calculated in two spatial positions for obtaining better results of its detection by artificial neural network. Recognition results are presented in the form of radar display. The neural network is tested on the intermediate positions of the mine from the reference training data. Approximation properties of the artificial neural network are investigated.