Research On Multi-class Motor Imagery EEG Classification Algorithm And Design Of Brain Computer Interface System

With the rapid development of science and technology,people's demand for intelligence has become more and more extensive.As a new type of human-computer interaction technology,Brain Computer Interface(BCI)has also come to people's attention from the laboratory and has a broad application prospect.But electroencephalogram(EEG)is a kind of non-linear and non-stationary signal,which is difficult to analyze.In order to improve the accuracy of EEG signal classification and the practicability of brain computer interface,this paper chooses multi task EEG signal classification for research,the difference between multi task and two classification is that the number of classifications is three or more and the difficulty of classification will increase.In addition,this paper also designs a brain computer interface system using EEG signal acquisition equipment.1.In the pre-processing stage of EEG signal: This paper introduces and uses band-pass filtering,Common Average Reference(CAR)and Independent Component Analysis(ICA),and to a certain extent eliminates the signals that unrelated to motor imagery EEG signal,thus providing a good basis for feature extraction and classification of EEG signal.2.In the phase of EEG signal recognition: Intrinsic Time-scale Decomposition(ITD)algorithm is introduced from the conventional signal decomposition algorithm,and the advantages of ITD are verified by comparing the decomposed signals;while selecting the decomposition component energy and AR model coefficient characteristics,phase synchronization is also introduced to describe the synchronization relationship between the electrodes during motion imagination,which makes up for the deficiency of the time-frequency feature in describing the spatial distribution feature;finally,Support Vector Machine(SVM)was selected to classify and identify the data of 12 subjects in the two types of data sets.The final classification accuracy was 88.99%,and the Kappa coefficient was 0.8517.The experimental results were analyzed and discussed,this method can effectively improve the classification results of multi class motor imaging EEG signals,and provides a good theoretical basis for the study of brain-computer interface technology.3.In the design of brain computer interface system,based on the above research,the brain computer interface system is developed.By wearing EEG acquisition equipment,the mouse on the computer can be controlled in real time,and then the character input and display can be realized by operating the mouse.The development of the system provides a reference for the application of BCI technology.