A single channel EMI signal separation method based on directly-mean empirical mode decomposition

ICA is a powerful decomposition method for time-domain series, except for the requirement that the number of observed signals and the source signals should be the same, which makes ICA fail to process single channel signals. In this paper, we propose a new method using directly-mean EMD, which is utilized to extract independent components from a single channel mixture. The proposed method could overcome the side effect of original EMD, and can be applied to the separation of EMI signals to locate interference sources. Simulation experimental results demonstrate the effectiveness of the proposed method, and show that the proposed method outperforms the comparison methods, such as the original EMD ICA and wavelet ICA.