Motor Intention And Motor Imagery EEG Data Analysis For Brain Computer Interface

Movement intention is a kind of command made by people to mobilize the brain resources related to movement when they are going to perform the action.Motor imagery is to use the mind to imagine the action,with the controller to achieve the subsequent actual action.Motor intention can be studied by motion-related cortical potentials and motor imagery can be studied by sensorimotor rhythms.The motor brain-computer interface not only helps patients control their bodies and realize self-care,but also helps patients recover their own motor ability.The study of motor intention and motor imagery is of great significance for motor brain computer interface.In this paper,the feature extraction method and classification recognition method of motion-related cortical potential and sensorimotor rhythm are studied.At the same time,the motor brain-computer interface based on EEG is realized.The main contents of this paper are as follows:1.Study of brain activity related to sports.Design intent and motion imagery experimental paradigm,the brain activity analysis related to sports in the sports related cortex potentials and sensorimotor rhythm,analyzing the relevant cortical potential experimental phenomenon and the differences in different actions,analyze sensorimotor rhythm potential experimental phenomenon and the differences in left and right side of the body,combining ERP analysis and statistical analysis of the experimental conclusions.2.According to different action sports intention recognition problem,most of the research on extracting only when cortex potentials characteristics relevant to the movement of time domain features,this paper extracted time and frequency domain characteristics of cortex potentials,relevant to the movement of the best individual feature selection algorithm based on mutual information feature selection,for each of the participants to choose the best dimension of feature selection,using the SVM classification,In the EEG data of this experiment,the average classification accuracy of upper and lower limb movements was 84%,and the average classification accuracy of upper limb and lower limb movements was 73%.3.To solve the problem of small sample EEG data in motor imagery,the common experiment of other subjects was introduced into the common space pattern algorithm of filter bank to generate regularized covariance matrix,so as to reduce the deviation caused by small sample number.In terms of feature selection,two feature selection methods were adopted,one was to select the appropriate regularization parameter pair,the other was to select the appropriate frequency band feature,and SVM was used for classification.In the EEG data of this experiment,the average classification accuracy of upper and lower limb experiments was 83%,and that of upper and lower limb experiments was 74%.4.In order to classify motor intention and motor imagery,a motor brain-computer interface based on EEG is designed in this paper.According to the motor intention in Chapter 3 and the feature extraction and classification of motor imagery in Chapter 4,a motor brain-computer interface based on EEG was designed.Spring Boot was used to build the background of the display module,and the front-end technology was used to build the system page.The EEG processing program was called in the background to display the classification and recognition results in the front-end page.