.png){kind=icon}
.png){kind=icon}
Return
Impulsive mode decomposition
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ymssp.2024.111227.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Pulse components commonly exist in natural signals and their extraction has received extensive concerns in many domains, such as machinery fault diagnosis, ECG denoising, non-destructive testing, chatter analysis, etc. Even though many signal decomposition methods (SDMs) have been applied to pulse component extraction, they are not originally tailored for pulse component extraction and cannot accurately and fully extract all pulse components. In this paper, impulsive mode decomposition (IMD) is originally tailored for adaptive pulse component extraction, and it can decompose a signal into impulsive modes and non-impulsive residual modes. This work mainly contributes three aspects: (i) A formal definition of impulsive mode; (ii) Geometrical mean-based pq-mean with four essential properties for quantification of impulsive modes; (iii) a novel iteratively-searching adaptive filterbank for extraction of impulsive modes. The effectiveness and ability of the proposed IMD are validated by a simulation case and three real-world application cases in machinery fault diagnosis and ECG signal denoising. Comparisons with variational mode decomposition, empirical wavelet transform, and minimum entropy deconvolution demonstrated the superiority of the IMD. The proposed IMD is promising to be used in various domains to extract pulse components of interest.
Keywords:
Impulsive mode
Signal decomposition
Adaptive filter
Sparsity measure
Minimum entropy deconvolution
Variational mode decomposition
Empirical wavelet transform
Journal
IF:
8.9
Papers:
1.3W
Citations:
6.6W