基于强化学习的雷达多域抗干扰策略生成技术  

Generation Technology of Radar Multi-Domain Anti-Jamming Strategy Based on Reinforcement Learning

作　　者：张连炜 董阳阳 李明 王贤铧 董春曦[1] 苏欣桐 ZHANG Lianwei;DONG Yangyang;LI Ming;WANG Xianhua;DONG Chunxi;SU Xintong(School of Electronic Engineering,Xidian University,Xi’an 710071,China;China Electronics Corporation Co.,Ltd.,Guilin 541001,China)

机构地区：[1]西安电子科技大学电子工程学院,西安710071 [2]中国电子信息产业集团有限公司国营第七二二厂,桂林541001

出　　处：《电子信息对抗技术》2024年第6期1-5,共5页Electronic Information Warfare Technology

基　　金：国家自然科学基金资助项目(61901332)。

摘　　要：雷达抗干扰能力的高低决定了其在复杂电磁环境下是否能正常完成探测、跟踪、制导等工作。针对雷达抗干扰问题,提出了基于强化学习的多域抗干扰策略生成技术。该方法通过对时域、频域、空间域和极化域进行优选,建立多域联合的雷达干扰-抗干扰规则库。通过强化学习得到动作价值矩阵,确定抗干扰策略,在极大降低后续决策复杂度的同时更能贴近战场情况,克服了强化学习算法难以适应大状态空间的问题。实验结果表明,与传统单一域的抗干扰决策相比,此方法可以较为准确地进行决策。Radar anti-jamming capability determines whether it can normally complete detection,tracking,guidance and other work in the complex electromagnetic environments.A multi-domain anti-jamming strategy generation technology based on reinforcement learning is proposed to address the issue of radar anti-jamming.A multi-domain joint radar jamming anti-jamming rule library is established by optimizing the time domain,frequency domain,spatial domain,and polarization domain.The action value matrix is obtained through reinforcement learning.The anti-jamming strategy is determined,which can greatly reduce the complexity of subsequent decisions while close to the battlefield situation.The problem is overcome that reinforcement learning algorithm is difficult to adapt to large state space.Experimental results show that compared with traditional anti-jamming decisions in a single domain,this method can make decisions more accurately.

关 键 词：雷达抗干扰 模板匹配 强化学习 抗干扰决策 

分 类 号：TN973[电子电信—信号与信息处理]