摘要
经验模态分解(empirical mode decomposition,EMD)以其自适应的滤波特性和多分辨率在很多非线性研究领域得到广泛应用,但在分解过程中常会出现虚假分量。针对此问题,基于信息论的知识提出利用K-L散度(Kullback-Leibler divergence,K-L)的虚假分量识别方法。该方法先将原始信号分解成若干个本征模态函数(intrinsic modefunction,imf),再分别计算原始信号与imf分量之间的K-L散度,然后将所求的K-L散度值从小到大排序,对应于较大K-L散度值的imf分量被视为虚假分量,可以去除。实验证明,该方法能够明显地区分出真实信号与虚假分量,准确而快速的得到信号的真实成分,消除虚假分量的影响。
Empirical mode decomposition （EMD） has been widely used in many nonlinear fields for its adaptive filtering properties and adaptive multiresolution, but there will be false components in the decomposition process. To solve this problem, a method of the false components identification of EMD based on KullbackLeihler divergence was proposed. First, the original signal was decomposed into several intrinsic mode functions （imf）; then the Kullback-Leibler divergence value between the original signal and each imf was calculated; finally the values were put in order. The imf, which corresponds to the big value, was regarded as a false component and can be removed. The experiment shows that this method can clearly distinguish the real components and the false ones, thus correctly and rapidly obtain the real composition of the signal and eliminate the influence of false components.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2012年第11期112-117,154,共6页
Proceedings of the CSEE
关键词
经验模态分解
K.L散度
虚假分量
信号处理
油膜涡动
empirical mode decomposition （EMD）
KullbackLeibler divergence
false component
signalprocessing
oil whirl
