基于半监督学习网络的雷达有源干扰识别

Radar Active Jamming Identification Based on Semi-supervised Learning Network

为提升具有少量标签的雷达有源干扰信号识别性能,提出一种基于半监督学习(SSL)网络的雷达有源干扰识别方法。首先,建立雷达有源干扰模型;然后,利用时频分析方法提取信号高维特征,并构建雷达有源干扰信号数据集;最后,开展基于FixMatch算法的雷达有源干扰识别。仿真结果表明,在数据集具有28个标签时模型精度为87.6%,具有175个标签时模型精度为92.8%,具有2800个标签时模型精度为94.7%。验证了SSL算法在具有少量标签的雷达有源干扰信号识别中有较好的效果。

In order to improve the performance of radar active jamming signal recognition with a few labels,a radar active jamming recognition method based on semi-supervised learning(SSL)network was proposed.Firstly,the radar active jamming model was established.Secondly,the time-frequency analysis method was used to extract the high-dimensional characteristics of the signal,and the radar active jamming signal data set was constructed.Finally,the identification of radar active jamming based on FixMatch algorithm was carried out.Simulation results showed that the model accuracy was 87.6% with 28 labels,92.8% with 175 labels and 94.7% with 2800 labels.It was verified that SSL algorithm has good effect on the identification of radar active jamming signal with a few labels.