Research On Ground Target Identification Technology Of Stepped-frequency Chirp Signal Radar

Stepped-frequency Chirp signal integrate the advantages of stepped-frequency signal and the linear frequency modulation signal,and has been widely used in target recognition with high range resolution.Based on the received target echo,a High Resolution Range Profile(HRRP)can be synthesized after information processing.HRRP can reflect the geometric structure and electromagnetic scattering characteristics of a target.Besides,HRRP has the advantages of easy acquisition and small data volume,so it is one of the important data sources for radar target identification.This paper mainly focuses on the research on ground target identification technology of the radar emitting stepped-frequency Chirp signal.The main contents are given as follows:1.Aiming at the problem of artifact in conventional HRRP imaging algorithm,an improved HRRP imaging algorithm for radar systems with stepped-frequency Chirp signals is studied in this paper.The algorithm first judges and weights to amplitude values that may have artifacts,and then splices the extracted information segments according to the order of sampling points to form HRRP.The experimental results shows that there is no artifact in the HRRP,obtained by the algorithm of this paper.2.Aiming at the problems of high feature dimension,high redundancy and noise sensitivity in the initial feature set,this paper studies a noise-robust feature selection algorithm based on HRRP.The algorithm fully considers the noise factor,uses the Relief F algorithm and the m RMR algorithm to comprehensively evaluate the features.Then,the sequence floating forward selection algorithm is employed to search the feature set to obtain the feature subset with better separability.The experimental results verify that the feature set selected by the feature selection algorithm proposed in this paper has the characteristics of good discrimination performance,low feature dimension,low redundancy and good noise robustness.3.The orientation sensitivity of HRRP reduce that there is a large number of other sample spaces within the boundary of the support vector data description(SVDD).However,other sample spaces will reduce the discrimination performance.To solve this problem,we investigate the radar target identification algorithm that combines the class algorithm and the SVDD algorithm.The algorithm first uses the density peak clustering algorithm to recluster the training samples.Then,the SVDD algorithm is introduced to train each cluster to obtain multiple decision boundaries.Experiments show that the radar target identification algorithm can effectively improve the discrimination performance and has strong noise robustness.4.Offline training system is one of the key steps to realize engineering application.Therefore,this paper uses the GUIDE to construct a target discriminator offline training system.The system includes the functional modules of training sample input,feature extraction,feature selection,discriminator training,and output training model parameters.Tests show that the system has the function of building a target discriminator model,with good interactivity and strong visualization.