基于自编码器的雷达信号联合预分选方法

An Autoencoder Based Method for Joint Pre-sorting of Radar Signals

将脉内特征提取、调制类型识别和聚类分选联合,提出了一种基于自编码器的雷达信号联合预分选方法。现有的基于脉内特征的聚类预分选方法需要预先设计特征参数提取方法,而所提方法可自动提取脉内特征参数,并根据聚类结果,对提取的特征参数进行调整,从而改变了以往分选算法的单向流程,引入了反馈机制以深入挖掘特征信息。仿真结果表明,该方法能在低信噪比环境下对雷达信号的脉内特征进行提取,并依靠脉内特征参数进行分选。

This paper combines intra-pulse feature extraction modulation type recognition with clustering sorting and proposes a joint pre-sorting method for radar signals based on the autoencoder.The existing clustering pre-sorting method based on intra-pulse features needs to design the feature parameter extraction method in advance while the proposed method can automatically extract the intra-pulse feature parameters and adjust the extracted feature parameters according to the clustering results which introduces a feedback mechanism to dig deep into the feature information and changes the one-way process of the existing sorting algorithm.The simulation results show that this method can extract the intra-pulse features of radar signals under low Signal-to-Noise Ratio(SNR)and use the intra-pulse feature parameters to sort.