Collective Artificial Intellegence Approach for the Problem of Object Classification with UWB GPR

Abstract

The problem of subsurface object detection and classification with the help of ultra-wideband ground penetrating radar and collective artificial intelligence is solving in this paper. The radar irradiates the ground by the short Gaussian electromagnetic impulse. The waves reflected from the models of ground and objects hidden in it are receiving by the system of four antennas. The 1Tx and 4Rx antenna system is used in this work. Electromagnetic wave radiation and propagation through the investigated volume is simulated by FDTD method. Obtained time dependences of electromagnetic field in four points of receiving are used as input data for the artificial neural networks. Four different neural networks simultaneously are making predictions of underground object presence, its type and position. Their answers on a number of testing cases serves as input data for the supreme neural network which is trained to make a final decision of classification result based on the history of the predictions of the four artificial neural networks. This system of classification is tested on noisy input data for different signal-to-noise ratios and on new object positions.