自动化学报2023，Vol.49Issue(12)：2544-2556,13.DOI:10.16383/j.aas.c201031
中值互补集合经验模态分解
Median Complementary Ensemble Empirical Mode Decomposition
摘要
Abstract
In order to restrain the mode splitting(MS)problem of the empirical mode decomposition(EMD)series methods,a median complementary ensemble EMD(MCEEMD)algorithm is proposed in this paper.We first present novel probabilistic tools to quantify the MS phenomenon of complementary ensemble EMD(CEEMD),which aims at demonstrating the effectiveness of using the median operator to replace the mean operator during the ensemble process.To combine the advantages of suppressing MS and residual noise,the MCEEMD algorithm integ-rates both median and mean operators within the ensemble process for the first time.Specifically,the MCEEMD al-gorithm is enlightened and featured by following procedures:1)Add N pairs of complementary white noise to the original signal to obtain 2N groups of intrinsic mode functions(IMFs)by EMD decomposition;2)By averaging each pair of the complementary IMFs,the 2N groups of IMFs are computed into N IMF groups;3)Assemble same-index components across the N groups of IMFs using the median operator to obtain the final IMFs within MCEEMD.Through typical simulations as well as two real-world cases,we show that the present work not only effectively alle-viates the MS problem,but also avoids two shortcomings of using a single median operator,i.e.,the poor decompos-ition completeness and the presence of burr in IMFs.关键词
模态分裂/中值算子/互补白噪声/互补集合经验模式分解Key words
Mode splitting(MS)/median operator/complementary white noise/complementary ensemble empirical mode decomposition(CEEMD)引用本文复制引用
刘淞华,何冰冰,郎恂,陈启明,张榆锋,苏宏业..中值互补集合经验模态分解[J].自动化学报,2023,49(12):2544-2556,13.基金项目
国家自然科学基金(81771928,62003298),云南省基础研究计划重点项目(202101AS070031),中国博士后科学基金资助项目(2020M683389)资助Supported by National Natural Science Foundation of China(81771928,62003298),Key Project of Fundamental Research of Yunnan Province(202101AS070031),and China Postdoctoral Science Foundation(2020M683389) （81771928,62003298）
