注意力机制和CNN结合的雷达辐射源个体识别  

Individual identification for a radar emitter combined with attention mechanism and CNN

作　　者：杨海宇 郭文普 康凯 何婧媛 边强[1] YANG Haiyu;GUO Wenpu;KANG Kai;HE Jingyuan;BIAN Qiang(College of Combat Support,Rocket Force University of Engineering,Xi’an 710025,China;College of Mathematics and Computer Science,Yan’an University,Yan’an 716000,China)

机构地区：[1]火箭军工程大学作战保障学院,西安710025 [2]延安大学数学与计算机科学学院,陕西延安716000

出　　处：《兵器装备工程学报》2023年第4期290-296,共7页Journal of Ordnance Equipment Engineering

摘　　要：针对雷达辐射源个体识别准确率低、抗噪性能差和训练时间长的问题,提出了一种注意力机制和卷积神经网络相结合的方法。首先根据雷达发射机功率放大器的硬件差异,建立雷达辐射源的系统模型;其次,对雷达信号进行双谱分析,得到的双谱图作为网络输入;然后,将注意力机制引入优化后的卷积神经网络,提高对个体特征的学习能力;最后,与现有方法对比,验证算法的有效性。实验证明,相比卷积神经网络,所提方法识别准确率提高5%,训练时间缩短一半。Aiming at the problems of low recognition accuracy,poor anti-noise performance and long training time of individual identification for a radar emitter,this paper proposes a method combining attention mechanism and convolutional neural network.Firstly,the system model of a radar emitter is established according to the hardware difference of a radar transmitter power amplifier.Secondly,the radar signal is analyzed by bispectrum,and the obtained bispectrum image is used as the input of the network.Then,the attention mechanism is introduced into the optimized convolutional neural network to improve the ability of individual identification.Finally,compared with the existing methods,the effectiveness of the algorithm is verified.The experimental results show that,compared with the convolutional neural network,the identification accuracy of the proposed method increases by 5%and the training time cuts in half.

关 键 词：注意力机制 卷积神经网络 雷达辐射源个体识别 双谱分析 

分 类 号：TN957.51[电子电信—信号与信息处理] TN974[电子电信—信息与通信工程]