基于时频图像分析的核爆与雷电电磁脉冲识别  

Discrimination of nuclear explosion and lightning electromagnetic pulse using time-frequency image analysis

作　　者：祁树锋[1] 李夕海[1] 韩绍卿[1] 陈蛟[1] 刘代志[1] 

机构地区：[1]第二炮兵工程学院,西安710025

出　　处：《强激光与粒子束》2013年第2期522-526,共5页High Power Laser and Particle Beams

基　　金：国家自然科学基金项目(40974037);中国博士后科学基金项目(201150M1554)

摘　　要：根据核爆和雷电电磁脉冲信号非平稳、非线性特点,对核爆电磁脉冲(NEMP)和雷电电磁脉冲(LEMP)信号进行了Hilbert谱分析,计算了二者Hilbert谱的图像区域特征,对二者进行了识别研究,并且从NEMP和LEMP不同的产生机理上对识别结果进行了分析。实验结果表明:以Hilbert谱的面积和重心,以及六维图像区域特征作为特征,对NEMP和LEMP的识别率达到了90%以上,可以对二者进行有效识别。in this paper, the Hilbert spectrum is used to analyze the nuclear explosion and lightning electromagnetic pulse for NEMP and LEMP signals based on their nomstationary and non-linear characters. The region features of Hilbert spectrum of NEMP and LEMP signals are calculated to discriminate them, and the discrimination result is analyzed based on different mecha- nisms of the nuclear explosion and lightning. The experiment results indicate that, using the nearest neighbor pattern classifica- tion method, the average discrimination rate of NEMP and LEMP signals is over 90% based on area, the center of gravity and six dimension image region features of Hilbert spectrum. Finally, we get a conclusion that the image region features of the Hilbert spectrum aIe effective features to discriminate NEMP and LEMP.

关 键 词：核爆电磁脉冲 雷电电磁脉冲 识别 时频图像分析 

分 类 号：TN913.325[电子电信—通信与信息系统]