Research On Brain-Computer Interface Technology Based On Motion Imagination

In recent years,the number of people with disabilities who have serious impairment of body motor function but still have healthy brains and can carry out complete thinking activities has been increasing.In order to help them achieve the rehabilitation of motor function,so that they can communicate smoothly with the external environment,Scholars in the society have begun to pay attention to the research of BCI technology based on motor imagery.brain-computer interface technology does not depend on the nervous system in the normal brain.It connects the brain directly to external devices,and converts an idea generated by the human brain into a corresponding control command to control the external devices without language and action.This brings hope to some special patient groups.EEG signals contain abundant physiological information and thinking information of patients.The analysis and research of such signals can accurately identify patients' thinking consciousness and understand their thoughts and intentions.This paper mainly analyzes the three aspects of denoising preprocessing,feature extraction and signal recognition classification of motor imagery EEG signals.Based on the analysis of ERD/ERS phenomena closely related to motor imagery EEG signals,the EEG signal is filtered by an 8~30Hz bandpass filter to realize the signal denoising preprocessing.In view of the problem that the CSP feature extraction method mainly deals with multi-lead EEG signals,and the useful signals in this paper are mainly concentrated on the three leads of C3,C4 and Cz,the method of combining R-CSP and OVO-CSP is proposed to extract the features of four different types of motor imagery EEG signals,which not only overcomes the problem of the small number of signal leads,but also improves the effectiveness of feature extraction.Aiming at the redundancy of the linear classification method based on Fisher criterion,a local statistical uncorrelated classification method based on weighted Fisher criterion(WLUDA)is proposed to classify EEG signals,which not only eliminates redundant information,but also achieves better recognition results than ordinary processing methods.Finally,the validity of the proposed methods is validated by the experiment of the motorimagery EEG signals collected in the laboratory.The experimental results show that the proposed signal processing methods effectively improve the recognition accuracy of EEG signals in motor imagery,and provides theoretical basis and technical support for the implementation of brain-computer interface system based on motor imagery and its application in daily life.