基于特征提取和机器学习的电磁目标识别方法  

Electromagnetic Target Recognition Method Based on Feature Extraction and Machine Learning

作　　者：徐洪志 姚家驰 刘超[2] 李彩霞[3] 蒋东翔[2] 孙腾龙 赵新青 XU Hongzhi;YAO Jiachi;LIU Chao;LI Caixia;JIANG Dongxiang;SUN Tenglong;ZHAO Xinqing(Unit 32142 of PLA, Baoding, Heibei 071000;Tsinghua University, Beijing 100084;Heibei University, Baoding, Heibei 071000)

机构地区：[1]中国人民解放军32142部队,河北保定071000 [2]清华大学,北京100084 [3]河北大学,河北保定071000

出　　处：《火控雷达技术》2022年第2期10-14,共5页Fire Control Radar Technology

摘　　要：电磁目标识别在军事领域中非常重要。为了准确识别电磁目标,本文提出一种基于特征提取和机器学习的电磁目标识别方法。首先,通过重采样、改变幅值和信号叠加三种方法扩充电磁目标数据库;然后,提取电磁目标的时域和频域统计特征;之后,通过主成分分析方法进行特征降维,保留前三个主成分;最后,用机器学习算法进行分类识别。研究结果表明,本文所提出的方法能够准确快速地识别电磁目标,在不同信噪比下的识别准确率均在98%以上。Electromagnetic target recognition is of great importance in the military field.In order to accurately identify electromagnetic targets,an electromagnetic target recognition method based on feature extraction and machine learning was proposed in this paper.Firstly,the electromagnetic target database was expanded after resampling,changing amplitude values and signal superposition;then,the time domain and frequency domain statistical features of the electromagnetic target were extracted;after that,principal component analysis was adopted to reduce the feature dimension and retain the first three principal components.Finally,classification and recognition were realized through machine learning algorithm.The research results showed that the proposed method can identify electromagnetic targets accurately and quickly,and the recognition accuracy is above 98%under different signal-to-noise ratios.

关 键 词：电磁目标 特征提取 特征降维 机器学习 

分 类 号：TN95[电子电信—信号与信息处理] TJ760[电子电信—信息与通信工程]