系统工程与电子技术 ›› 2024, Vol. 46 ›› Issue (3): 898-905.doi: 10.12305/j.issn.1001-506X.2024.03.15

• 传感器与信号处理 • 上一篇    下一篇

基于双胞循环神经网络的雷达捷变频行为识别

孟宪鹏1, 刘利民1,*, 董健1, 王力1,2, 胡文华1   

  1. 1. 陆军工程大学石家庄校区电子与光学工程系, 河北 石家庄 050003
    2. 中国人民解放军32203部队, 陕西 华阴 714200
  • 收稿日期:2022-07-14 出版日期:2024-02-29 发布日期:2024-03-08
  • 通讯作者: 刘利民
  • 作者简介:孟宪鹏(1986—), 男, 工程师, 博士研究生, 主要研究方向为雷达对抗、人工智能
    刘利民(1971—), 男, 教授, 博士, 主要研究方向为电子对抗、雷达信号处理
    董健(1982—), 男, 讲师, 博士, 主要研究方向为雷达处理、机器学习
    王力(1994—), 女, 助理工程师, 硕士研究生, 主要研究方向为雷达对抗
    胡文华(1970—), 男, 副教授, 博士, 主要研究方向为雷达处理、雷达性能检测与故障诊断
  • 基金资助:
    国家自然科学基金(61571043)

Radar frequency agility behavior recognition based on bi-cell recurrent neural network

Xianpeng MENG1, Limin LIU1,*, Jian DONG1, Li WANG1,2, Wenhua HU1   

  1. 1. Department of Electronic and Optical Engineering, Army Engineering University Shijiazhuang Campus, Shijiazhuang 050003, China
    2. Unit 32203 of the PLA, Huayin 714200, China
  • Received:2022-07-14 Online:2024-02-29 Published:2024-03-08
  • Contact: Limin LIU

摘要:

雷达程控捷变频行为具有一定的抗窄带瞄准式干扰能力, 同时能够实现测量和动目标指示等功能, 给干扰引导带来一定的困难。对此,提出随机频率模板的方法, 对雷达程控捷变频行为进行建模, 并设计了一种双胞循环神经网络识别程控捷变频行为。仿真实验结果表明, 双胞循环神经网络能够有效识别雷达程控捷变频行为, 并以一定的概率预测未来的频率序列, 能够有效地为窄带瞄准式干扰提供引导。仿真结果也表明, 所提网络能够有效记忆和识别一组非线性时间序列。

关键词: 捷变频, 行为识别, 循环神经网络, 记忆细胞

Abstract:

Radar program-controlled frequency agility behavior has a certain ability to resist narrow-band aiming jamming, and can realize the functions of measurement and moving target indication, which brings some difficulties to jamming guidance. For this, a random frequency template method is proposed to model the program-controlled frequency agility behavior of radar, and a bi-cell recurrent neural network (BRNN) is designed to identify the program-controlled frequency agility behavior. The simulation results show that the BRNN can effectively identify the frequency agility behavior of radar program-controlled, and predict the future frequency sequence with a certain probability, which can effectively provide guidance for narrow-band aiming jamming. The simulation results also show that the proposed network can effectively remember and identify a group of nonlinear time sequence.

Key words: frequency agility, behavior recognition, recurrent neural network (RNN), memory cell

中图分类号: