一种基于聚类和集成学习的混合脉冲去交错方法  

A Deinterleaving Method for Mixed Pulse Signals Based on Clustering and Ensemble Learning

作　　者：刘建 柴新新 刘正成 LIU Jian;CHAI Xin-xin;LIU Zheng-cheng(The 8th Research Academy of CSSC,Yangzhou 225101,China)

机构地区：[1]中国船舶集团有限公司第八研究院,江苏扬州225101

出　　处：《舰船电子对抗》2022年第1期77-80,共4页Shipboard Electronic Countermeasure

摘　　要：在现代复杂电磁环境下,来自不同辐射源的信号在时域、空域和频域高度重叠交错。传统的脉冲去交错方法在应对相似脉冲参数方面的表现不尽人意。提出了一种利用聚类和集成学习解决问题的方法。该方法利用脉冲的多个参数进行聚类,通过所提取的能够表达脉冲序列趋势的特征对集成学习模型进行训练,利用训练好的模型对聚类的信号组群进行判断,是否属于同一个辐射源。对所提方法进行仿真的结果表明该方法能有效地进行混合脉冲去交错。In modern complex electromagnetic environment,pulse signals from different emitters are highly overlapped and interlaced in time domain,spatial domain and frequency domain.The tradi-tional pulse deinterlacing method does not perform satisfactorily in dealing with similar pulse pa-rameters.This paper proposes a method employing clustering and ensemble learning to solve this issue.This method utilizes multiple parameters of the pulse to cluster pulse signals into small groups,trains an ensemble learning model by means of the extracted features which can represent pulse sequences’variation trend,and uses the trained model to judge whether the clustered signal groups belong to the same emitter.The simulation results of the proposed method show that the method can effectively deinterlace mixed pulses.

关 键 词：去交错 聚类 集成学习 

分 类 号：TN971.1[电子电信—信号与信息处理]