摘要
随着电子技术的发展,雷达信号调制方式趋于复杂化,对雷达信号波形进行可靠的识别是雷达侦察系统面临的一个难题。其中,类线性调频雷达信号(LFM、Frank码、P1~P4码)是识别难度较大的一类信号。针对该问题,论文提出了一种基于深度学习的雷达波形自动识别方法,首先计算类线性调频雷达的基于S-method的时频分布,在此基础上计算信号时频分布的Zernike矩,然后通过卷积神经网络进行分类识别。实验表明,在-2dB信噪比下,信号的整体识别率达到94.33%,能够有效识别雷达波形。
With the development of electronic technology,the modulation mode of radar signal tends to be complicated.It is a difficult problem to identify the waveform of radar signal reliably.Among them,LFM radar signal(LFM,Frank code,P1-P4 code)is a kind of signal which is difficult to identify.Firstly,the S-Method based time-frequency distribution of the LFM radar is calculat⁃ed,and then the Zernike moment of the time-frequency distribution of the signal is calculated,and then it is classified and recog⁃nized by convolutional neural network.The experimental results show that the overall recognition rate of the signal reaches 94.33%at-2dB signal to noise ratio.
作者
夏沭涛
金堃
付宇鹏
于文龙
XIA Shutao;JIN Kun;FU Yupeng;YU Wenlong(Institute of Information Fusion,Naval Aviation University,Yantai 264001;School of Basis of Aviation,Naval Aviation University,Yantai 264001;No.91213 Troops of PLA,Yantai 264001)
出处
《舰船电子工程》
2024年第2期96-99,142,共5页
Ship Electronic Engineering
基金
国家自然科学基金项目“跨域场景下海上舰船目标综合识别关键技术研究”(编号:61271499)资助。
关键词
类线性调频雷达信号
时频分布
ZERNIKE矩
卷积神经网络
linear frequency modulated radar signal
time-frequency distribution
Zernike moment
convolutional neural network
