Research On BCI System Based On EEG Exercise And Imagination To Regulate Emotions Online

Brain-computer interface(BCI)is a revolutionary human-computer interaction technology,which aims to record specific neural activities,construct cognitive models,decode users' intentions and psychological states,and develop specific closed-loop neural feedback system.Among them,Emotion Brian computer interface(E-BCI)is a closed-loop emotion computing system that integrates detection and feedback.The research of E-BCI in emotion monitoring is of great significance because the long-term accumulation of negative emotions will lead to emotional disorders,which will affect physical and mental health,and even lead to brain diseases.However,at present,few studies use a motor imagery strategy to regulate emotion,ignoring the effect of cognitive intervention on emotion.Therefore,based on Electroencephalogram(EEG),this paper mainly studies the method of subject-independent emotion recognition,explores the appropriate motor imagery paradigm,and initially develops an online emotion recognition system to monitor the effect of motor imagery strategy on emotion regulation.The main contents of this paper are as follows:(1)This paper explores whether the experimental paradigm based on EEG imagery motor parameters can use Support vector machine(SVM)and Extreme learning machine(ELM)algorithms for pattern classification.The experimental results show that the experimental paradigm of EEG imagery motor parameters can effectively induce the Event Related Desynchronization/Event-Related Synchronization(RED/ERS)phenomenon,and SVM and ELM algorithm can effectively identify the imagery motor patterns.(2)To solve the problems of the time-domain,frequency-domain,time-frequencydomain information is not fully utilized in emotion recognition,and the classification model is subject-dependent.In this paper,we extract all kinds of features commonly used in emotion recognition and use the Relevant Features(Relief F)feature selection algorithm to sort all kinds of features.Then,we select the feature training classifier with higher weight,and compare the classification performance of SVM,Linear Discriminant Analysis(LDA),K-Nearest Neighbors(KNN),Random Forest(RF),Deep Belief Network(DBN)algorithm,and propose a subject-independent emotion recognition framework.Then,based on the subject-independent emotion recognition framework,a preliminary online emotion recognition system is developed.(3)This paper designed the experimental paradigm of motor imagery and emotion stimulation,and preliminarily studied the regulating effect of motor imagery on emotion from the perspective of the neural model.The results show that motor imagery can promote positive emotions.Motor imagery strategy may provide a new reference for emotion regulation.