一种水声弱信号奇异熵特征提取方法  

Research on extraction method of underwater acoustic singular entropy features

作　　者：叶博源 梁喆 刘文帅 吕孟婷 宋建强 YE Bo-yuan;LIANG Zhe;LIU Wen-shuai;LU Meng-ting;SONG Jian-qiang(Dalian Scientific Test and Control Technology Institute,Dalian 116013,China)

机构地区：[1]大连测控技术研究所,辽宁大连116013

出　　处：《舰船科学技术》2022年第18期155-158,共4页Ship Science and Technology

摘　　要：本文提出一种基于奇异值熵的固有模态函数筛选重组方法,将经验模态分解得到的固有模态函数经过信息熵计算筛选,去除低信噪比信号中具有较高能量的环境噪声,提升舰船辐射噪声特征的信噪比,从而一定程度上改善对舰船目标的远距离识别效果。实验结果表明,经过信息熵筛选后的固有模态函数重构信号明显突出了舰船辐射噪声信号的原有特征,可有效提高对远距离目标的识别率。This paper proposes a method for screening and recombination of intrinsic mode functions based on singular value entropy. The intrinsic mode functions obtained from empirical mode decomposition are filtered through information entropy calculations to remove high-energy environmental noise in low signal-to-noise ratio signals, and improve The signalto-noise ratio of the ship’s radiation noise characteristics can improve the long-distance recognition of ship targets to a certain extent. The experimental results show that the inherent modal function reconstructed signal after information entropy screening clearly highlights the original characteristics of the ship’s radiated noise signal, which can effectively improve the recognition rate of long-distance targets.

关 键 词：舰船目标识别 经验模态分解 HILBERT-HUANG变换 奇异值熵 BP神经网络 

分 类 号：TN911.7[电子电信—通信与信息系统]