EEG Signal Classification Method Based on Feature Priority Analysis and CNN

In recent years, with the advancement of technology, more and more people with disabilities use prosthetic limbs to compensate for the function of normal upper limbs, so the demand for the upper limb movement recognition has increased. In addition, with the development of brain-computer interfaces, a large number of academic institutions are dedicated to identifying the upper limb movements through EEG signals. In this paper, a new EEG signal classification method is proposed to identify the upper limb movements. Firstly, the importance of EEG detection electrodes is sorted by random forest algorithm, and the higher priority electrodes are screened out. Then the wavelet transform method is used to extract the features of the original data that have been screened out. After that, the EEG signals are classified according to the extracted features by using convolutional neural network. The model designed in this paper has a classification accuracy rate of 93.22% on the WAY-EEG-GAL dataset, which is an effective EEG signal classification model.