Research On EEG-Based Emotion Recognition And Regulation System

Emotion is a physiological and psychological phenomenon produced by the human body to adapt to changes in the environment,and it is a reflection of people’s attitude and behavior towards objective things.Brain-Computer Interface(BCI)is a new information and communication technology that can convert signals from the human brain directly into computer commands to complete communication with external machines without relying on machinery or the central nervous system of the human brain.Combining emotion and BCI provides new ideas for the development of emotion,and the main applications currently include emotion recognition brain-computer interface and emotion modulation brain-computer interface.Although emotion recognition brain-computer interface has made certain development,it suffers from insufficient feature extraction,slow speed,poor accuracy of emotion recognition,and is difficult to be applied to real-time emotion classification.The development of emotion regulation brain-computer interface is relatively slow,and no clear regulation index has been formed,so that emotion regulation cannot be performed for specific situations.In response to the above problems,this paper investigates the mechanism of emotion recognition and regulation,extracts features from the collected emotion signals and classifies them online,and the recognized signals can be used for online regulation,and develops an EEG-based emotion recognition and regulation system platform.The platform built in this paper is divided into two parts: offline experimental platform and online experimental platform.The offline platform can be used for emotion elicitation acquisition and offline recognition analysis,and the online platform can be used for online feature extraction and classification as well as emotion regulation.The emotion elicitation part includes the establishment of an emotion video library and signal acquisition,which can improve subjects’ attention and the quality of the collected EEG signals by watching videos.The offline recognition analysis part utilizes a deep learning algorithm to more fully extract EEG features and improve the emotion recognition effect.The online platform includes an online feature extraction and classification part and an emotion regulation part: The online feature extraction and classification part can read EEG data in real time,perform qualitative and quantitative analysis of EEG data,and extract its features to build a multidimensional classification model that intuitively reflects the brain activity.The emotion regulation part can use both positive ideation therapy and motor imagery therapy based on motor observation to regulate the emotion for the purpose of restoring the emotional state identified.In this paper,the brain activity was monitored and analyzed in real time,the signal characteristics of different emotional states were studied,and different deep learning algorithm models were used for classification,among which the CNN-LSTM-Attention model analyzed and recognized the experimental data of four different emotions,and its recognition accuracy could reach 89%,and the classification effect of emotion recognition was significantly improved.The EEG signals in the negative emotion state and after the two methods of emotion regulation were compared,both of them could improve the preference of attention to external emotional stimuli in healthy people,and make the depressed people pay less attention to external negative stimuli and more attention to positive stimuli,thus improving the ability of emotion regulation.In summary,the system has good recognition,feedback and regulation functions,which lay a solid foundation for subsequent clinical trials and applications.