摘要
针对非合作侦察接收机只在降噪后才能开展后续检测识别工作的问题,结合降噪自编码器和生成对抗网络的优势,构建噪声增强网络与信号增强网络进行对抗训练。噪声增强网络往带噪信号中掺杂更复杂的噪声分量,信号增强网络则是尽可能地降低带噪信号中的噪声分量。二者在对抗训练的过程中,噪声增强网络生成复杂高维噪声的能力和信号增强网络降噪的能力都在提升。训练结束后,信号增强网络具备更好的降噪性能,为后续检测识别工作降低难度。
Aiming at the problem that the non-cooperative reconnaissance receiver can carry out subsequent detection and recognition only after noise reduction,combining the advantages of noise reduction self encoder and generating countermeasure network,the noise enhancement network and signal enhancement network are constructed for countermeasure training.The noise enhancement network adulterates more complex noise components into the noisy signal,and the signal enhancement network reduces the noise components of the noisy signal as much as possible.In the process of confrontation training,the ability of noise enhancement network to generate complex high-dimensional noise and the ability of signal enhancement network to reduce noise are improving.After the training,the signal enhancement network has better noise reduction performance,which reduces the difficulty of subsequent detection and recognition.
作者
曹鹏宇
杨承志
石礼盟
吴宏超
CAO Pengyu;YANG Chengzhi;SHI Limeng;WU Hongchao(School of Air Operations and Services,Aviation University of Air Force,Changchun 130022,China;Unit 93671 of the PLA,Nanyang 474350,China)
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2021年第9期2493-2500,共8页
Systems Engineering and Electronics
基金
国防科技卓越青年基金(315090303)资助课题。
关键词
低截获雷达
生成对抗网络
降噪自编码器
信号增强
low probability of intercept(LPI)radar
generative adversarial networks(GAN)
denoising autoencoder
signal enhancement
