EEG signal recognition based on wavelet transform and neural network

For the problem that EEG signal is difficult to identify, this paper proposed the method of wavelet transform combined with neural network for signal processing and feature extraction of EEG signal. The original EEG data was processed by wavelet de-noising method, the delta wave, theta wave, alpha wave and beta wave signal are obtained by db5 wavelet decomposition. Then the energy value of sub-band signal as the characteristic vector of the EEG signal is identified. The BP and RBF neural network are designed to identify the EEG signal in the condition of fatigue, sober and concentration. The experimental results show that the correct rate of RBF is 92% and BP is 75%, and the mean square error of RBF identification is generally less than BP neural network.