节点文献

基于希尔伯特—黄变换的水声瞬态信号检测方法研究

Research on the Method of Underwater Acoustic Transient Signal Detection Based on Hilbert Huang Transform

【作者】 薛飞

【导师】 梁国龙;

【作者基本信息】 哈尔滨工程大学 , 水声工程, 2012, 硕士

【摘要】 水声瞬态信号检测一直以来都是信号处理领域的难点和重点。一方面水声环境复杂、信噪比低的特点给信号的检测带来不便。另一方面,瞬态信号往往包含丰富的目标特征信息,对它的检测、估计和识别具有重大的价值,因此长久以来备受国内外学者的青睐。文章首先简要介绍了几种常见的瞬态信号检测手段,浅析了Hilbert-黄变换的提出和研究概况,从经验模态分解和Hilbert谱两方面系统地介绍了Hilbert-黄变换的理论基础,仿真实现了多分量信号的模态分解和Hilbert谱分析。在此基础上介绍了基于经验模态分解(EMD)的一种加噪信号处理方式:集总经验模态分解(EEMD)。重点研究了经验模态分解、集总经验模态分解特性,讨论了Hilbert-黄变换的完备性、正交性、频率分辨率和边界处理等问题,指出了EMD的不足,仿真说明了EEMD并没有从根本上解决模态混叠问题,并重点分析了模态分解的二进制滤波器特性。然后,本文基于瞬态信号的数学模型,提出了基于EEMD、信号重构、瞬时能量密度级和的三种低信噪比下瞬态信号检测方法,通过绘制接收机工作特性曲线(ROC)将后两种方法与传统能量检测器性能进行对比。最后通过湖试数据和海试数据处理验证了本文方法的可行性和有效性。

【Abstract】 Underwater transient signal detection is always the emphasis and difficulty in the signalprocessing field. The one hand, the complex underwater acoustic environment and low SNRput signal detection into inconvenience. On the other hand, transient signals always contain awealth of target signature information, which have great significance to the signal detection,estimation and recognition. So it is the favour of domestic and foreign scholars for so long.Firstly, some typical transient signal detection methods were introduced briefly, and theproposition and the research survey of Hilbert-Huang Transform were analysed. From the tworespects of empirical mode decomposition and Hilbert spectrum, the theory of Hilbert-HuangTransform was introduced systematically, and the mode decomposition and Hilbert spectralanalysis of multi-component signals simulations was achieved. Then one type of pre-whitedprocessing method based on Empirical Mode Decomposition (EMD) was introduced, which isEnsemble Empirical Mode Decomposition (EEMD). The characteristics of EMD and EEMDwere mainly studied, including completeness, orthogonality, frequency resolution ratio,boundary etching, and so on. The deficiency of EMD was pointed out. It was showed bysimulation that the mode mixing haven’t been resolved radically by EEMD. The binary filtercharacteristics of mode decomposition were analysed emphatically.Then, based on the mathematic model of transient signal, three methods were proposedwhich were found on EEMD, Hilbert-Huang Transform reconstruction and the energy densitylevel. By drawing the receiver operating characteristic (ROC) curve, the detectionperformances of last two methods were compared with the traditional energy detection’s.Finally, the feasibility and validity of the methods in this paper were verified by processingthe lake and sea trial data.

  • 【分类号】TN911.23
  • 【被引频次】5
  • 【下载频次】493
节点文献中: 

本文链接的文献网络图示:

本文的引文网络