EEG classification based on improved Genetic Algorithm channel selection

The control of intelligent peripheral devices such as robot arm and mouse movement through brain-computer interface needs to be based on accurate feature extraction and analysis of EEG signals. The signal-to-noise ratio of EEG signals is low and the signal dimension is high. The feature extraction of multi-channel EEG signals will introduce interference signals, so it is necessary to eliminate noise and redundant information signals by channel selection algorithm. Aiming at the problem that the traditional channel selection method needs to pre-define the number of selected channels and cannot meet individual differences, this paper proposes an EEG signal classification method based on improved Genetic Algorithm for channel selection. Then, Discrete Wavelet Transformation, and Common Spatial Pattern are used to extract the features of motor imagery EEG signal data after channel selection, and the EEG signal features are classified by Support Vector Machines. The experimental results show that the average classification accuracy of EEG signals after channel selection is 5% higher than that of all channels, and the problem of individual differences in the number of channels and channels is solved.