基于深度学习的同型雷达的脉冲中频信号分离技术  

Intermediate Frequency Pulse Signal Separation Technology of Homogeneous Radar Based on Deep Learning

作　　者：曾令华 高由兵[1,2] 梁超 赵耀东 史小伟 ZENG Linghua;GAO Youbing;LIANG Chao;ZHAO Yaodong;SHI Xiaowei(Southwest China Research Institute of Electronic Equipment,Chengdu 610036,China;National Key Laboratory of Electromagnetic Space Security,Chengdu 610036,China)

机构地区：[1]中国电子科技集团公司第二十九研究所,成都610036 [2]电磁空间安全全国重点实验室,成都610036

出　　处：《电子信息对抗技术》2023年第6期53-60,共8页Electronic Information Warfare Technology

基　　金：国家自然科学基金资助项目(U20B2070)。

摘　　要：传统电子战接收机系统提取中频信号脉冲描述字的过程会造成大量信息丢失,导致在信号分离过程中难以适应越来越复杂的低信噪比电磁环境。基于深度学习模型,直接建立从中频采样信号到分离信号特征表达的映射,避免了中间过程的信息损失和误差累积。通过以标记数据进行训练的方式提供专家知识,该模型得以使用基于有监督学习的优化方法,实现高准确率的分离。实验结果表明,该方法相对基于传统脉冲描述字的方法在低信噪比环境下进行信号分离的场景中具有很大的优势。The process of extracting intermediate frequency signal pulse descriptor from traditional electronic warfare receiver system will cause a lot of information loss,which makes it difficult to adapt to the increasingly complex electromagnetic environment with low signal-to-noise ratio in the process of signal separation.Based on the deep learning model,the mapping is directly established from intermediate frequency sampling signal to separate signal feature expression.The information loss and error accumulation are avoided in the intermediate process.Expert knowledge provided through training with tagged data.The optimization method based on supervised learning is used by this model to achieve high accuracy of separation.The experimental results show that this method has great advantages over the traditional pulse descriptor method in the scene of signal separation in low signal to noise ratio environment.

关 键 词：雷达信号分离 人工智能 信号去交错 中频信号处理 

分 类 号：TN971[电子电信—信号与信息处理]