Research On Emotional Recognition Technology Based On Brain Electrical Signals

Emotion is a psychological-physiological process,judging and evaluating emotions has been integrated into every aspect of our lives,helping us to achieve better development in many areas.The theme of this paper is to identify the emotion around the brain electrical signal,and to study the channel selection and feature extraction methods involved in this process,the goal is to obtain better accuracy of emotion recognition.This paper focuses on the processing process of brain electrical signals,feature extraction,feature selection,channel selection,and final emotional classification algorithm,and test ingress with training and verification on the exposed DEAP emotional data set,achieves up to 94.2%.The first is the pre-processing of data,the paper uses power spectral density and cost calculations to determine the choice of data channels,not only can eliminate redundant characteristics,but also greatly reduce the data dimension and data size,convenient after the extraction of features.For feature extraction,the paper uses the empirical modal decomposition algorithm and sample entropy combination for feature extraction,the former is used to break down the brain electrical signal,its adaptive corresponding to solve the nonlinear characteristics of brain electrical signals,as for the sample entropy algorithm,it can solve the fluctuation characteristics of brain data.The two work together to achieve a good effect.Finally,we detail the characteristic classification method and use SVM for emotional classification identification.For the analysis of the final result,the paper focuses on three variables in feature extraction,namely,the sample data segment size N,the similar tolerance r and embedded dimension m,and analyzes their effect on the accuracy of the result according to the results,so as to obtain the superior parameter combination.Then the results were compared with the results of other research papers using DEAP data sets for emotional classification,and found that the accuracy improved a lot and had good scientific research value.