基于独立分量分析的EMD模态混叠消除方法研究.pdfVIP

学兔兔 第33卷 第7期 仪 器 仪 表 学 报 VoL 33 No．7 2012年7月 Chinese Journal of Scientific Instrument Ju1．2012 基于独立分量分析的EMD 模态混叠消除方法研究 汤宝平，董绍江，马靖华 (重庆大学机械传动国家重点实验室 重庆 400030) 摘 要：针对经验模态分解(empirical mode decomposition，EMD)算法中存在的模态混叠现象，提出一种应用独立分量分析进行模 态混叠消除的新方法。首先应用形态学滤波方法将信号中的噪声予以消除，减少由于噪声因素造成的模态混叠现象。然后将信 号进行EMD分解，获得不同的IMF分量，将存在模态混叠成分的IMF分量进行相空间重构，利用基于峭度最大化的独立分量分解 算法实现混叠成分的分离，仿真分析和工程应用的结果表明，所提方法能有效地消除EMD分解过程中的模态混叠现象。 关键词：经验模态分解；模态混叠；形态学滤波；相空间重构；独立分量分析 中图分类号：TP206 ．3 文献标识码：A 国家标准学科分类代码：460．40 Study on the method for eliminating mode mixing of empirical mode decomposition based on independent component analysis Tang Baoping，Dong Shaojiang，Ma Jinghua (The State Key Laboratory ofMechanical Transmission，Chongqing University，Chongqing 400030，China) Abstract：Aiming at the existing problem of mode mixing in empirical mode decomposition(EMD)，a new method based on independent component analysis(ICA)is proposed．Firstly，the morphological filtering method is used to remove noise and reduce the mode mixing caused by noise factor．Then，the EMD algorithm is used to decompose the signal into different IMF components，and the IMF components that have the problem of mode mixing are selected and reconstructed in the phase space．Finally，the ICA algorithm based on the kurtosis maximization is adopted to assume the independence among the sources and achieve the separation of the overlapped components．The resuhs of simulation analysis and engineering application show that the proposed method can effectively eliminate the mode mixing in EMD． Key words：empirical mode decomposition(EMD)；mode mixing；morphological filtering；phase space reconstruc— tion；independent component analysis(ICA) 但是