基于经验模式分解的广义互相关时延估计  

Generalized Cross-correlation Time-delay Estimation Based on EMD

作　　者：孙书学[1] 顾晓辉[1] 吕艳新[1] 

机构地区：[1]南京理工大学机械工程学院,江苏南京210094

出　　处：《探测与控制学报》2009年第2期5-9,13,共6页Journal of Detection & Control

摘　　要：提出一种基于经验模式分解(Empirical Mode Decomposition,EMD)的广义互相关时延估计方法,用来解决信号非稳态且采集信号的各个通道间存在相关性噪声的条件下如何提高时延估计精度的问题。利用EMD处理非稳态、非线性信号具有良好的自适应性的特点,将声信号分解为多个本征模态函数(Intrinsic Mode Function,IMF),然后对相应的IMF进行互相关获得多尺度时延值。针对多个时延值不相等的问题,给出了选择精确时延的频谱一致性和时延矢量匹配两个准则,提高了该方法的实用性;并将这一理论应用于声阵列对声源的定位。试验证明:该方法能够在信号非稳态且采集信号的各个通道间存在相关性噪声条件下实现时延的精确估计,并提高了定位精度和时延估计的稳健性。A new method for generalized cross-correlation time-delay estimation （TDE） based on EMD （empirical mode decomposition） is proposed, which can improve the accuracy of TDE in the case that the signals are non-stationary and there are correlative noises among different channels. EMD is applicable and self-adaptive to nonlinear and non-stationary signal processing, which can decompose the acoustic signals into several IMFs （intrinsic mode function）. Then the corresponding IMFs from different channels are cross-correlated to obtain multi-scale TDEs. In view that the TDEs mentioned above are not equal, two criterions are given to select accurate TDE which are spectrum consistency and vector matching, improving the practicability of the proposed method. The method is also applied to the localization of acoustic source with acoustic array. Test results show that the proposed method can gain accurate TDE and improve the localization accuracy and the robustness of TDE in the case that the signals are non-stationary and there are correlative noises among different channels.

关 键 词：经验模式分解 本征模态函数 声阵列 时延估计 

分 类 号：TB535[理学—物理]