基于经验模态分解的物体入水声检测及测向研究  

Research on detection and azimuth estimation for the splash sound for target's water entry based on empirical mode decomposition

作　　者：李关防 赵安邦[1] 惠俊英[1] 王逸林 

机构地区：[1]哈尔滨工程大学水声工程学院,黑龙江哈尔滨150001

出　　处：《声学学报》2008年第6期548-554,共7页Acta Acustica

基　　金：水声技术国防科技重点实验室基金(9140C200101080C2005);哈尔滨工程大学基础研究基金(HEUFT07072)资助项目

摘　　要：物体入水声是一种瞬态信号,其波形由击水声和若干气泡脉动组成。传统的矢量信号处理方法对此类瞬态信号的检测和测向会出现困难,尤其是在信噪比较低时检测不到入水声信号。经验模态分解是一种突出信号局部瞬态特性的非线性分析方法,将矢量传感器接收的声压、振速信息分解为不同的固有模态函数,利用文中提出的模态声强器的方位估计算法,可以实现瞬态信号的检测和测向。湖试和海试结果表明该方法能把本地干扰和入水声分解到不同的模态函数中,利用模态声强器可以在本地强干扰下有效检测到入水声信号出现的时间,并可以实现测向。The splash sound of target＇s water entry is an instantaneous signal, consisting of impact component and bubble components. It is difficult to detect and estimate the azimuth of this signal by traditional azimuth estimation methods using a single vector sensor, especially when the SNR is low. Empirical mode decomposition is a nonlinear analysis method which can emphasize signals＇ instantaneous characteristic. In this paper, the pressure and particle velocities are decomposed into different intrinsic mode functions （IMF）, and the MAIA （mode acoustic intensity averager） method can realize the instantaneous signals＇ detection and azimuth estimation. The lake experiment and sea trial results show that this method can decompose water-entry sound and ambient interference into different IMFs, so as to detect the starting time of water-entry sound signal and estimate azimuth effectively.

关 键 词：经验模态分解 声检测 测向 物体 固有模态函数 非线性分析方法 水声信号 矢量传感器 

分 类 号：TB566[交通运输工程—水声工程]