[Research of movement imagery EEG based on Hilbert-Huang transform and BP neural network]

This paper introduces the characteristics of the Hilbert-Huang transform (HHT), and studies the classification of movement imagery EEG based on the HHT method and BP neural network. After preprocessed, the movement imagery EEG data were descomposed with empirical mode decomposition (EMD) into a series of intrinsic mode functions (IMFs). Then the low frequency IMFs were removed, and the rest of IMFs were conducted by Hilbert transform to get Hilbert marginal spectrum. The marginal spectrum subtracted values between the channal C3 and channal C4 were selected as the original features which were then decreased the dimension by the principal components analysis so as to be jointed with EEG complexity to construct the feature vector. The BP neural network was utilized to classify the EEG pattern of left and right hand motor imagery. The brain computer interface (BCI) competition II data set III was selected to carry out the discrimination, and the classification accuracy rate is up to 87.14%, which is a comparably good result and proves HHT to be a feasible and effective method on EEG analysis.