Study On Radar HRRP Target Recognition Based On Neural Network

Radar target recognition refers to using radar equipment to obtain information of characteristics from targets,and uses priori knowledge from all sorts of targets which have been mastered to recognize properties of unknown targets.At present, the main techniques of pattern recognition applied on radar target recognition include statistical pattern recognition methods,fuzzy pattern recognition methods and neural network identification methods.Because the neural network owns the abilities of self-adaptive,self-organization,self-learning,so it can build memory by learning sample,which can declare unknown model to be the closest sub-memory and that is also the reason why it is applied widely in pattern recognition and in the field of radar target identification.This paper has researched on high-resolution radar target recognition deeply by extracting high-resolution radar target echo signals based on characteristics and combining neural network technology.Firstly, this paper has introduced neural network technology and the basic theory of high-resolution radar target signal detection.Secondly, we have analyzed on high-resolution radar target echo signals and have pointed out three characteristics of the echo signals which is orientation sensitivity, shifting the sensitivity and amplitude sensitivity. On this basis, This paper has proposed extracting the signal information entropy value as the echo signal characteristic values that are combined with other traditional features of the information of echo signals to be multi-dimensional feature vector, and have proposed the methods of high resolution radar target recognition based on BP neural network and RBF neural network combined with neural networks.High-resolution radar echo signal pre-processing,multi-feature value extraction, rational using BP neural network and RBF neural networks,have improved the rate of recognition effectively. Finally, we have proposed a new neural network-combined neural network to carry on high-resolution radar target identification.The experimental results have proved the effectiveness of the neural network technology fully.