Research On Fatigue Driving Detection And Early Warning Technology Based On EEG Signals

With the acceleration of the pace of daily life in the world,more and more cars are driving fast on the road,and the road traffic safety issues are becoming more and more prominent.According to effective data,in various kinds of traffic accidents,traffic accidents caused by fatigue driving have reached 25% to 30%,accounting for 83% of traffic accident mortality.As a result,driving safety issues have attracted widespread attention worldwide.Driver fatigue status is hidden and individual differences.Direct detection method cannot be used to judge driver fatigue status,and the subjective judgment of driver has little effect on fatigue status,so the potential harm of fatigue driving is extremely great.Therefore,if the driving state of the driver can be detected in a timely and effective manner as well as a early warning is used in some way,this will be an urgent task for those drivers who are endangered by fatigue driving,and it will also be a research topic theoretical value and practical significance for preventing traffic accidents.As the “gold standard” for detecting fatigue,EEG signals have the advantages of objectivity and accuracy.In order to solve the current problems of fatigue driving detection and early warning technology based on EEG signals,the main research contents include:(1)Aiming at the problems of accuracy and individual differences in the fatigue driving detection process,a number of simulated driving experiments are performed by multiple people to reflect the personalized experimental design.Then a relatively complete experimental scheme is determined.And finally,this scheme is adopted to carry out the fatigue driving simulation experiment based on 64 lead electrode and single electrode.(2)For the non-real-time fatigue driving data,an EEG analysis method combining time-frequency and nonlinear dynamics is used.The wavelet packet decomposition is used to process the EEG signal,calculate the energy value of each rhythm,and obtain the fatigue index threshold.Then,reconstruct the decomposed EEG signal,use the sample entropy to perform nonlinear dynamic analysis,and obtain the sample entropy threshold.Refer to the fatigue index threshold and sample entropy threshold to divide different degrees of fatigue.(3)For the real-time fatigue driving data,concentration and relaxation,power spectral density,and blink frequency are used to analyze and obtain thresholds.Then,divide different fatigue levels according to thresholds.In addition,compare KNN,ANN,and SVM classification algorithms,and finally use KNN pairs Fatigue classification.Besides,the three fatigue indicators are synthesized using an improved DS evidence theory synthesis algorithm,and the results show that the algorithm's judgment accuracy rate for different fatigue levels is slightly higher than that of single feature judgment fatigue.(4)Design and develop an IPv6-based fatigue driving early warning platform.The platform development language uses Java and HTML,and the database uses My Sql.It mainly implements real-time fatigue status detection,real-time positioning,driving status monitoring,and fatigue status early warning.Big data analysis is used to fully explore the relationship between driver driving status and fatigue,to achieve individualized management of individual drivers,and to effectively improve the real-time and accuracy of fatigue detection.This article is based on the “CERNET Innovation Project”(No.NGII20170712)set up by Network Center of China Education and Research Network and Cernet Coporation.This paper focuses on the detection and early warning of fatigue driving based on EEG signals,and obtains the comprehensive fatigue index threshold according to the actual situation,which provides reliable indicators for EEG to judge fatigue driving and assist objective detection of fatigue driving.In addition,an IPv6-based fatigue driving early warning platform is designed and developed to form a complete fatigue driving early warning system.