Selection Method of IMFs Based on Noise-Assisted MEMD and Relevant Information

Brain-computer interface (BCI) is a technology to control external equipment through electroencephalogram (EEG), but also EEG is often detected along with noise. The filtered noise is very important for the application of brain-computer interface. Multivariate empirical mode decomposition (MEMD) can decompose a multi-channel nonlinear and non-stationary time series into a series of physically meaningful intrinsic modal component (IMF) components. However, in many frequency identification algorithms based on MEMD, the reconstruction of the signal is only a simple screening of the signal-dominated IMF component, but the useful information in other noise-dominated IMF components is ignored. In this paper, an IMFs selection method based on noise-assisted MEMD and relevant information and the application of this method in the SSVEP brain-computer interface is studied, among them, canonical correlation analysis (CCA) algorithm is used for feature extraction. The experimental results show that the accuracy of this method is 3.6-5.3% higher than that of the classic scheme and it has higher stability (standard deviation less than 4%), which provides a new method for the application of SSVEP brain-computer interface.