Research On Ground Moving Target Identification Based On Ground Motion

The seismic signal that is generated from the movements of ground targets can be acquired by single axis acceleration sensors.This thesis designs the algorithm of target recognition to realize the identification of target after the analysis and feature extraction of the captured signal.It is of great practical value to identify the targets using the ground motion signals in the field of runway intrusion prevention and safety monitoring applications,such as airport perimeter and area protection.Firstly,the mechanism of the generation and propagation of seismic waves is studied,and the principle of ground motion acceleration sensor is introduced.The seismic signals that is generated by ground moving targets are easily affected by environmental noise,so the wavelet packet transform method are employed to reduce the noise;Then,it decompose the reduced noise signal using the empirical wavelet transform method.Compared with the empirical mode decomposition method,this method has some prominent advantages,such as less decomposed modes and faster calculation speed.At the same time,this paper extracts the time domain,frequency domain parameters from the seismic signal after reduced noise and extracts the energy indicators from the intrinsic mode function(IMF)that is selected according to the cross-correlation rule.Finally,these parameters are combined to construct the feature vector as the input of the multiple class support vector machine for training and prediction.Aiming at the problem that the traditional support vector machine parameters are difficult to choose,the genetic algorithm are used to find the optimal parameters including the penalty factor C and the kernel parameters g of the support vector machine(SVM).The experimental results show that the genetic algorithm optimization support vector machine(GA-SVM)model can improve the performance of classifier.Moreover,the genetic algorithm optimized support vector machine model is superior to the unimproved cross validation support vector machine model in training speed and recognition efficiency.The GA-SVM model can correctly identify the ground moving targets such as personnel,vehicles and aircraft.