Improvement of ensemble Empirical Mode Decomposition by a band-limited white noise

Abstract

The empirical mode decomposition (EMD) was a tool proposed for analysis of nonlinear and non-stationary signals and was successfully used in various applications. However, one of the major drawbacks of the EMD is the appearance of mode mixing when the signal consists of widely disparate scales. The EEMD was developed to solve the mode-mixing with the assistance of added white noises which produced the residue noise in the signal reconstructed. The effect of the added white noise can be reduced to a negligibly small level by increasing the number of ensemble trials of a few hundred, to the detriment of increased computing time. A modified EEMD, termed MEEMD, was proposed to improve the EEMD computational efficiency substantially by replacing the white noise by a band-limited noise. But, the MEEMD did not specify the type of filter used and its characteristics. In this study, we showed that the right choice of the filter type and its order improved appreciably the computational efficiency of the MEEMD.