Median ensemble empirical mode decomposition

Xun Lang; Naveed Ur Rehman; Yufeng  Zhang; Lei Xie; Hongye Su

doi:10.1016/j.sigpro.2020.107686

Median ensemble empirical mode decomposition

Xun Lang, Naveed Ur Rehman, Yufeng Zhang^*, Lei Xie^*, Hongye Su

^*Corresponding author for this work

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Journal article › Research › peer-review

Abstract

Ensemble empirical mode decomposition (EEMD) belongs to a class of noise-assisted EMD methods that are aimed at alleviating mode mixing caused by noise and signal intermittency. In this work, we propose a median ensembled version of EEMD (MEEMD) to help reduce the additional mode splitting problem of the original EEMD algorithm. That is achieved by replacing the mean operator with the median operator during the ensemble process. Our use of the median operator is motivated by a rigorous analysis of mode splitting rates for both EEMD and MEEMD. It is shown that EEMD comes with irremovable new mode splitting while the proposed method can greatly reduce this problem on a breakdown point of 50%. This work is verified by extensive numerical examples as well as industrial oscillation case in terms of reducing the mode splitting.

Original language	English
Article number	107686
Journal	Signal Processing
Volume	176
Number of pages	8
ISSN	0165-1684
DOIs	https://doi.org/10.1016/j.sigpro.2020.107686
Publication status	Published - Nov 2020
Externally published	Yes

Keywords

Empirical mode decomposition
Ensemble empirical mode decomposition
Median
Mode splitting

Access to Document

10.1016/j.sigpro.2020.107686

Cite this

@article{040ddff59c9c451fac5e7605772be6d6,

title = "Median ensemble empirical mode decomposition",

abstract = "Ensemble empirical mode decomposition (EEMD) belongs to a class of noise-assisted EMD methods that are aimed at alleviating mode mixing caused by noise and signal intermittency. In this work, we propose a median ensembled version of EEMD (MEEMD) to help reduce the additional mode splitting problem of the original EEMD algorithm. That is achieved by replacing the mean operator with the median operator during the ensemble process. Our use of the median operator is motivated by a rigorous analysis of mode splitting rates for both EEMD and MEEMD. It is shown that EEMD comes with irremovable new mode splitting while the proposed method can greatly reduce this problem on a breakdown point of 50%. This work is verified by extensive numerical examples as well as industrial oscillation case in terms of reducing the mode splitting.",

keywords = "Empirical mode decomposition, Ensemble empirical mode decomposition, Median, Mode splitting",

author = "Xun Lang and Rehman, {Naveed Ur} and Yufeng Zhang and Lei Xie and Hongye Su",

year = "2020",

month = nov,

doi = "10.1016/j.sigpro.2020.107686",

language = "English",

volume = "176",

journal = "Signal Processing",

issn = "0165-1684",

publisher = "Elsevier",

}

TY - JOUR

T1 - Median ensemble empirical mode decomposition

AU - Lang, Xun

AU - Rehman, Naveed Ur

AU - Zhang, Yufeng

AU - Xie, Lei

AU - Su, Hongye

PY - 2020/11

Y1 - 2020/11

N2 - Ensemble empirical mode decomposition (EEMD) belongs to a class of noise-assisted EMD methods that are aimed at alleviating mode mixing caused by noise and signal intermittency. In this work, we propose a median ensembled version of EEMD (MEEMD) to help reduce the additional mode splitting problem of the original EEMD algorithm. That is achieved by replacing the mean operator with the median operator during the ensemble process. Our use of the median operator is motivated by a rigorous analysis of mode splitting rates for both EEMD and MEEMD. It is shown that EEMD comes with irremovable new mode splitting while the proposed method can greatly reduce this problem on a breakdown point of 50%. This work is verified by extensive numerical examples as well as industrial oscillation case in terms of reducing the mode splitting.

AB - Ensemble empirical mode decomposition (EEMD) belongs to a class of noise-assisted EMD methods that are aimed at alleviating mode mixing caused by noise and signal intermittency. In this work, we propose a median ensembled version of EEMD (MEEMD) to help reduce the additional mode splitting problem of the original EEMD algorithm. That is achieved by replacing the mean operator with the median operator during the ensemble process. Our use of the median operator is motivated by a rigorous analysis of mode splitting rates for both EEMD and MEEMD. It is shown that EEMD comes with irremovable new mode splitting while the proposed method can greatly reduce this problem on a breakdown point of 50%. This work is verified by extensive numerical examples as well as industrial oscillation case in terms of reducing the mode splitting.

KW - Empirical mode decomposition

KW - Ensemble empirical mode decomposition

KW - Median

KW - Mode splitting

UR - http://www.scopus.com/inward/record.url?scp=85086444130&partnerID=8YFLogxK

U2 - 10.1016/j.sigpro.2020.107686

DO - 10.1016/j.sigpro.2020.107686

M3 - Journal article

SN - 0165-1684

VL - 176

JO - Signal Processing

JF - Signal Processing

M1 - 107686

ER -

Median ensemble empirical mode decomposition

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this