Research Of Detection Method On Fatigue Driving State Based On Chaos Characteristic Of ECG

Fatigue driving is a very dangerous driving behavior,which causes a large number of traffic accidents every year.Therefore,in order to prevent fatigue driving,many scholars have carried out a lot of research on fatigue driving state detection,which has become a research hotspot.Research shows that physiological signals are objective and do not change with subjective consciousness,so using physiological signals to detect drivers’ fatigue driving state has higher accuracy.In the research status of detecting driver’s fatigue state based on physiological signal,it can be found that many scholars mainly judge the fatigue driving state according to the change law of physiological signal index parameters with the deepening of fatigue,and rarely can give a quantitative threshold to judge whether the driver is in the fatigue state.This paper presents a method to determine the threshold of fatigue driving state based on the chaos characteristics of ECG.The main work includes:(1)The chaos characteristics of ECG signals in fatigue state and non fatigue state are analyzed.This paper takes the ECG data in the Drozy database of the University of Liege in Belgium as the driving data source of this study.Then,the complexity,sample entropy and fuzzy entropy of ECG data are extracted,and the variation characteristics of three chaos eigenvalues under fatigue and non fatigue conditions are analyzed.The results show that complexity,sample entropy and fuzzy entropy can characterize the driver’s fatigue driving state,which lays a foundation for determining the threshold of fatigue driving state.(2)A method to obtain the threshold of fatigue driving state by principal component analysis and ROC curve analysis is proposed.Through the correlation analysis of the three chaos parameters,it is found that there is correlation and redundant information,so the principal component analysis method is used to reduce the dimension of data and establish the comprehensive index of fatigue driving.Then the ROC curves of the three chaos characteristic parameters and the comprehensive index of fatigue driving are analyzed and compared.The results show that the comprehensive index of fatigue driving not only improves the sensitivity and specificity of a single chaos characteristic parameter,but also reduces the misjudgment rate.Finally,according to the principle that the maximum value of Youden index is the best threshold,the threshold of fatigue driving is determined.(3)The simulation experiment of driving fatigue state is carried out based on ECG signal to verify the accuracy of ROC curve method in obtaining fatigue state threshold.Firstly,the changes of fatigue degree before and after the experiment were analyzed in detail by means of subjective questionnaire,and the ECG signal was denoised.Then three chaos characteristic parameters that can characterize the fatigue driving state are extracted,and the threshold of fatigue driving state is determined by principal component analysis and ROC curve method.Finally,by comparing the test set with the threshold value,the method of obtaining the threshold value of ROC curve and the method of obtaining the threshold value by calculating the peak and valley threshold formula of comprehensive index T are verified and compared.The research results show that the method of obtaining the fatigue threshold value by ROC curve proposed in this paper has better recognition effect,and the accuracy reaches 96%.