EEG Signal Research Based On Emotional Evocation

One of the most important ways to communicate with human beings is to identify their emotions.With the development of brain science,neuroscience and computer technology,human beings are no longer satisfied with the emotional communication between people.Human beings have begun to explore human-computer interaction,hoping that machines can correctly identify emotions and assist people to make decisions quickly and accurately.In recent years,in the biomedical field,the research of emotion recognition based on EEG signal has always been a research hotspot,and many excellent recognition models and analysis methods have emerged,but there is still a lot of room for improvement in practicability and applicability.This paper is based on the analysis and classification of emotional stimulation EEG.Based on the in-depth study of EEG and emotion related theories,a feature extraction method based on variable scale symbol compensation transfer entropy is proposed,and the acquisition channel is selected and optimized in order to reduce the amount of calculation and improve the real-time performance.The main contents of this paper are as follows:(1)In order to solve the limitation of single channel EEG emotion recognition,according to the bidirectional transmission mechanism of brain neural information,the information interaction between different channels is analyzed.In this paper,linear Granger causality and nonlinear transfer entropy method are used to qualitatively analyze the information interaction between channels during emotional stimulation.At the same time,HOG is introduced to extract the features of the generated relationship matrix graph to reduce redundant features and improve the classification accuracy.(2)In order to solve the effect of transitive entropy on feature extraction,the variable scale symbol compensation transfer entropy is proposed in this paper.The principle of the method is analyzed in detail,and how to compensate for the instantaneous causality is explained.The traditional transfer entropy algorithm and the improved variable scale symbol compensation transfer entropy algorithm are combined with HOG to extract the features of the relationship matrix between stress and calm state,and SVM is used to compare the classification accuracy of the two features.The results showed that the classification accuracy of VSSCTE is improved to 96.74% without increasing the operation time.Finally,by comparing with the research results at home and abroad,the superiority of VSSCTE proposed in this paper for EEG signal feature extraction is confirmed.(3)On the basis of theoretical analysis,this paper considers the scene of practical application,VSSCTE is used to construct brain network,by studying their correlation and causality,the network measurement is analyzed to explore the topological structure of EEG causal brain network,so as to understand the cooperative working mode of different brain regions in emotional stimulation.Then the Relief F feature selection algorithm is used to select the EEG channel.Under the premise of ensuring the accuracy of classification,the computation time is reduced and the experimental difficulty is reduced.At the same time,considering that different subjects may have different optimal channels,in order to verify the universality of the selected channels,this paper predicts the emotional accuracy of EEG signals of different subjects.The results showed that except for the two subjects whose classification accuracy rate was low(still greater than 90%),the classification accuracy of other subjects reached a high level,the best classification accuracy rate reached 97.71%,and the average classification accuracy rate was 94.36%,which confirmed that the optimized channel has real-time and universal applicability in EEG emotion analysis.