摘要
针对当前常用的基于信号时频域的雷达辐射源个体识别方法无法完整表征雷达个体特征,难以满足复杂电磁信号环境以及不同雷达工作模式的雷达个体识别需求问题,在信号时频分布特征基础上,通过分析计算信号的时域、频域、变换域等不同域特征,综合构建了雷达辐射源多域特征库,提出最大相关最小冗余多域特征优选方法,创新设计了基于多域特征融合的双路输入神经网络识别模型,并使用注意力机制自适应地关注不同信号环境下雷达重点特征权重,解决仅使用时频域进行时频分布识别率较低的问题。通过实侦实验分析对比,验证了该方法可以较好地适应同时多信号应用场景,个体识别准确率提升了12%以上。
The common radar emitter individual identification methods based on the time-frequency domain of the signal fail to fully characterize the individual characteristics of radar,and they are far away from meeting the needs of radar individual identification in the complex electromagnetic signal environments and different radar working modes.In view of the above,this paper analyzes and calculates the characteristics of the signal on the time-domain,frequency-domain,and transformation domain on the basis of the time-frequency distribution characteristics of the signal,and then comprehensively constructs a multi-domain feature database of radar emitters.A method for optimizing the maximum relevance and minimum redundant(MRMR)multi-domain features is proposed.A dual-input neural network recognition model based on multi-domain feature fusion is designed innovatively,and the attention mechanism is used to adaptively pay attention to the weights of key radar features in different signal environments,so as to eliminate the low recognition rate of time-frequency distribution when only the time-frequency domain is used.After the analysis and comparison of experiments on real time reconnaissance,it is verified that the proposed method can adapt to the simultaneous multi-signal application scenarios satisfactorily,and the accuracy rate of individual recognition is improved by more than 12%.
作者
毛秀华
樊昀
郑瑾
王强
MAO Xiuhua;FAN Yun;ZHENG Jin;WANG Qiang(Beijing Institute of Tracking and Telecommunications Technology,Beijing 100094,China;National Key Laboratory of Space Integrated Information System,Beijing 100094,China)
出处
《现代电子技术》
北大核心
2025年第17期21-28,共8页
Modern Electronics Technique
关键词
信号细微特征
时频分布
最大相关最小冗余
注意力机制
双输入神经网络
个体识别
subtle feature of signal
time-frequency distribution
MRMR
attention mechanism
dual-input neural network
individual recognition
作者简介
毛秀华(1982-),女,江苏泰州人,硕士研究生,副研究员,研究方向为遥感信息智能处理;樊昀(1976-),男,江西鹰潭人,博士研究生,研究员,研究方向为遥感信息智能处理。
