Target Recognition Based on Seismic Sensors and Neural Network

With the development of sensing technology and the applications of data fusion, acquisition and processing of reconnaissance information enters a new period. On the basis of research on information acquisition, features extraction, classification and decision of target in battlefield, acquiring features of seismic signal of target by ground seismic sensors, in virtue of neural network, pattern recognition of typical military target based on seismic signal has been solved. It is very important that to select which network to be the structure of training, it will effect to the precision and the time length of the pattern recognition directly, and it has important meaning to pattern recognition in the battlefield. Aiming to the shortage of the BP network on the foundation of the Elman type network structure, a new structure design of IRN network with biased units was presented, to, experiments prove it can be used in target recognition and it is superior than traditional methods.