The Research Of Psychologicla Stress Modeling Based On EDA Signals

Psychological stress has gradually become an important factor affecting the health of individuals.To effectively regulate and control psychological stress,it is necessary to understand and judge the intensity of psychological stress.The electrodermal activity signal is a non-invasive measurement index,which can easily capture changes in the body's mood,is easy to obtain,and is not subjectively controlled by the individual.It is a good physiological indicator for assessing the state of psychological stress.This paper proposes a method based on single-modal multi-level psychological stress state recognition.It mainly uses the real driving psychological stress database published by MIT and the electrodermal activity data collected by the psychological stress inducing experiment designed by the research team,and screens out the effective electrodermal activity signals as the original samples of the real environment and the laboratory environment,respectively.These samples were used for pretreatment,feature extraction,feature optimization and classification identification of electrodermal activity signals.The Fisher projection algorithm is combined with Naive Bayesian algorithm and support vector machine algorithm to optimize the characteristics of the electrodermal activity and classify the psychological stress state.At the same time,based on the work of the MIT multimedia laboratory,the contribution of four unique characteristics of the electrodermal activity to the classification results was studied.The results show that the recognition rate of Naive Bayes classifier in psychological stress recognition is better than the support vector machine algorithm in the above two datasets;based on the real environment data,plus four unique characteristics of the electrodermal activity signals.The psychological stress recognition rate is significantly better than the recognition rate without including these four features.At the same time,the optimal projection space of feature optimization based on Fisher projection algorithm is also studied.The results show that the 2D feature space is the optimal feature subset space.An identification model based on the psychological stress state of the electrodermal activity signal was initially established.In the model of psychological stress state recognition in real environment and laboratory environment,the optimal recognition rate is 81.82%,and the highest recognition rate for low-intensity psychological stress state is 95.45%.It is indicated that three different psychological stress states based on the electrodermal activity signals can also achieve better recognition results.Compared with multiple physiological signals,or based on a single physiological signal to identify whether there are two kinds of psychological stress states,the multi-level psychological stress recognition model based on single physiological signals is not only simple,but the performance of the model is even better than some.There is a model,which can make a good balance between the performance of the model and the computational load,and this is a promising development direction of the actual personal psychological stress state level recognition instrument in the future.