基于AP-DBSCAN聚类的弹道目标进动特征提取  

Procession Feature Extraction of Ballistic Targets Based on AP-DBSCAN Clustering Algorithm

作　　者：陈蓉[1] 冯存前[1,2] 王义哲 许丹 

机构地区：[1]空军工程大学防空反导学院,西安710051 [2]信息感知技术协同创新中心,西安710077

出　　处：《弹箭与制导学报》2017年第3期109-113,共5页Journal of Projectiles,Rockets,Missiles and Guidance

基　　金：国家自然科学基金(61372166)资助

摘　　要：进动是弹道目标识别的重要特征。以锥体弹头为研究对象,文中提出了一种基于宽带雷达组网的锥体目标进动特征提取方法。首先建立弹道目标进动模型,利用AP聚类算法,根据回波信号的强度进行初步聚类,然后通过DBSCAN算法,剔除噪声点,将非噪声信号分类并求平均值。在此基础上,分别估计出不同雷达体制下各散射中心的幅、相信息,进而解算出弹道目标的进动参数。仿真结果表明,在信噪比较小的情况下,目标的进动参数估计精度仍较高。Precession is a critical feature in the identification of ballistic targets. Aimed at cone-shaped warhead, a method to extract the precession feature of cone-shaped target, which is based on the mixed netted radars consisting of both narrowband and wideband ones, is proposed. Firstly a precession model of ballistic target is developed, and the AP clustering algorithm is introduced to cluster the echo signals on the basis of the signal intensity. Noisy points are deleted through the DBSCAN algorithm and the average values of non-noisy signals are calculated after the classification. Based on the work mentioned above, the amplitude and phase information of each scattering center in different netted radars is estimated. Then the precession parameters of ballistic targets are calculated. Simulation results validate that the precision of estimation on the precession parameters stays on high accuracy under the condition of a low signal noise ratio （SNR）.

关 键 词：宽带雷达 AP聚类 DBSCAN密度聚类 进动特征提取 

分 类 号：TN957[电子电信—信号与信息处理]