部分集成局部特征尺度分解:一种新的基于噪声辅助数据分析方法

摘要

Local characteristic-scale decomposition(LCD)was a new non-stationary data analysis mehod,which was proposed recently and similar to empirical mode decomposition(EMD).In order to solve its mode mixing problem,firstly a noise-assisted data analysis method named ensemble local characteristic-scale decomposition(ELCD) is presented.However,since ELCD inherited the shortcomings of ensemble empirical mode decomposition(EEMD)and complementary ersemble empirical mode decomposition(CEEMD),in the same time,based on the new randomicity detecting method-permutation entropy(PE),another method for restraining mode mixing called partly ensemble local characteristic-scale decomposition(PELCD)had been proposed in this paper.Lastly,the novel method was compared with the existing method(CEEMD) by analyzing simulation data and real data and the results indicate that the proposed method can restrain the phenomenon of mode mixing effectively and is superior to ELCD and other traditional noise-assisted method in aspects of inhibiting false components and improving the accuracy of components.%局部特征尺度分解(Local Characteristic-Scale Decomposition,LCD)是最近提出的一种类似于经验模态分解(Empirical Mode Decomposition,EMD)的非平稳信号分析方法.为解决LCD方法的模态混淆问题,论文首先提出了基于噪声辅助分析的集成局部特征尺度分解方法(Ensemble LCD,ELCD).然而,ELCD有类似于总体平均经验模态分解(Ensemble EMD,EEMD)和互补总体平均经验模态分解(Complementary,CEEMD)的固有缺陷,在此基础上,同时结合最近提出的随机性检测方法——排列熵(Permutation Entropy,PE),论文提出了部分集成局部特征尺度分解(Partly Ensemble LCD,PELCD)方法.仿真数据分析表明,论文提出的PELCD方法不仅能够有效地抑制LCD分解的模态混淆,而且在抑制伪分量的产生以及分量精确性等方面要优于CEEMD和ELCD方法.