基于并联深度学习网络的雷达有源干扰智能识别方法  

Intelligent Recognition Method of Radar Active Jamming Based on Parallel Deep Learning Network

作　　者：姜正云 舒汀[1] 何劲[1] 郁文贤[1] JIANG Zhengyun;SHU Ting;HE Jin;YU Wenxian(Shanghai Key Laboratory of Intelligent Sensing and Recognition,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海交通大学上海市智能探测与识别重点实验室,上海200240

出　　处：《现代雷达》2021年第10期9-14,共6页Modern Radar

基　　金：国家自然科学基金资助项目(61771302)。

摘　　要：针对传统的雷达有源干扰识别方法存在特征参数对干扰样式敏感,识别准确率不高等问题,提出了一种基于深度学习的雷达有源干扰智能识别方法,设计了一种残差网络(ResNet)和长短时间记忆网络(LSTM)相并联的新型网络结构。该方法基于多维度信息联合处理,可提高干扰识别的稳健性。通过外场试验,对常规的6种雷达有源干扰样式进行识别性能验证,识别准确率达到94.80%,证明了该文的方法具有较好的工程应用前景。The traditional radar active jamming recognition is a pattern recognition method by manually extracting multiple feature parameters and sending them to a predetermined classifier for classification. This method has some problems, such as sensitive characteristic parameters to jamming patterns and not high recognition accuracy, etc. In response to the above problems, an intelligent recognition method for radar active interference based on deep learning is proposed in this paper, and a new network structure in which a residual network and a Long short-term memory network are connected in parallel is designed. In addition, this method is based on the joint processing of multi-dimensional information, which can improve the robustness of interference recognition. The recognition performance of six kinds of conventional radar active jamming patterns is verified by field experiments, and the recognition accuracy rate reaches 94.80%, which proves that the proposed method has a good engineering application prospect.

关 键 词：残差网络 长短时间记忆网络 并联网络 雷达有源干扰识别 实测数据验证 

分 类 号：TN972[电子电信—信号与信息处理]