Research And Implementation Of Psychological Stress Recognition Algorithm Based On HRV

The fast pace and fierce competition in modern society make everyone face the stress of work,study and life.Different degrees of psychological stress will have a certain impact on people.If the degree of psychological stress is too large or long-term accumulation,it will cause mental or physiological diseases in the human body,so,Effective identification and accurate evaluation of psychological stress is a hot topic in recent years.At present,the evaluation of psychological stress is mainly through questionnaire survey,but this method is subjective and not accurate,and using physiological signals to identify psychological stress is more objective and reliable.Therefore,this paper proposes to use heart rate variability(HRV)to identify and evaluate the state of psychological stress in human body,and to analyze the HRV index which can reflect the activities of cardiac autonomic nerve,so as to realize the recognition and evaluation of psychological stress more effectively.In this paper,the psychological stress recognition algorithm based on HRV is studied.According to the different psychological stress states,the psychological stress inducing experiment is designed through the mental arithmetic task.The ECG signal is collected by ECG acquisition circuit and NI myRIO embedded development platform.The signal is stored,pre-processed,feature extracted and classified.In this paper,the pressure sample data set is constructed,and the wavelet transform algorithm is used to filter and detect the ECG signal.Then the RR interval is calculated by the detected R wave to obtain the HRV signal,and the characteristic parameters of HRV which can represent the pressure degree are extracted.Among them,statistical method is used to extract time domain features,autoregressive model(AR)method is used to extract frequency domain features to overcome the shortcomings of spectral aliasing,low resolution and so on.Poincare scatter graph method is used to extract nonlinear features with a total of 20 dimensions.On this basis,the features are compared and analyzed.In this paper,the classification model of psychological stress degree is established,and the BP network model and support vector machine(SVM)classifiers are constructed to meet the experimental requirements.Because the dimension of the feature space sent to the classifiers is too high,Therefore,a psychological stress recognition algorithm based on particle swarm optimization(PSO)algorithm combined with classifiers is proposed to overcome the"dimensional disaster" of the performance degradation of classifiers due to excessive dimensions.In addition,in order to effectively improve the recognition rate of psychological pressure,an improved PSO optimization algorithm is proposed,which introduces the PSO model with contraction factor to cancel the boundary limit of speed.The parameter optimization of BP neural network and SVM is realized by selecting appropriate parameters to ensure the convergence of PSO algorithm.The basic PSO algorithm and the improved PSO algorithm are combined with the two classifiers respectively to obtain the recognition rate of PSO-SVM algorithm,PSO-BP algorithm,improved PSO-SVM algorithm and improved PSO-BP algorithm to psychological pressure.Finally,using the mixed programming of LabVIEW and MATLAB,the function modules such as data acquisition,data storage,data processing,feature extraction,psychological stress recognition,pressure recognition level display and so on are designed to realize the psychological stress evaluation system.The experimental results show that the recognition rates of PSO-SVM algorithm,PSO-BP algorithm,improved PSO-SVM algorithm and improved PSO-BP algorithm to psychological stress are 82.50%,84.50%,90.17%,94.83%,respectively.The recognition rate of the improved PSO optimization classifier is significantly higher than that of the basic PSO algorithm,which shows that the improved PSO algorithm has a strong generalization ability in the optimization.At the same time,the recognition rate of the improved PSO-BP algorithm is 4.66%higher than that of the improved PSO-SVM algorithm,which shows that the improved particle swarm optimization algorithm can significantly improve the classification accuracy of the BP neural network.Therefore,the research and evaluation system design of psychological stress recognition algorithm provides objective and effective basis and means for psychological stress evaluation and intervention,which has certain social significance.