Study On Driver Risk Characteristics In Extra-long Highway Tunnels Based On Physiological And Behavioral Indicators

Due to the particularity of geometry and driving environment in highway tunnels,the light and dark adaptation process at the exit and entrance section will cause the physiological and psychological behavior to change and thus affect the vehicle operation according to “perception-cognitive-response” process.This situation will further increase the dynamic driving risk and the possibility of traffic accidents.Human factors could be the important and direct causes of traffic accidents.The risk from drivers is comprised of individual risk attributes and risk driving state.Therefore,how to reveal the risk driving state in the driving process using a real-time,continuous and comprehensive way by measuring the physiological indicators and behavior indicators of the driver when they drive through the special road section exerts essential application values.These values are reflected in the intervention of driving behavior,the development of driving assistance system,the design of road geometry,the layout of safety facilities.Except for the special circumstances,such as fatigue,alcoholism driving and so on,the risk driving state includes both the mental workload and the behavior risk state.The thesis aims to study diver risk characteristics in extra-long highway tunnels based on physiological and behavioral indicators.The detailed contents include the analysis of driver risk state and the method of identifying the risk attributes of existing drivers.At the same time,the effects of the driving experience on the indicators characteristics and the model results were considered.A total of 30 drivers(including 15 skilled drivers and 15 unskilled drivers)were selected to conduct naturalistic driving experiments in two typical tunnel environments.The general characteristics of the physiological indicators and speed of these two types of drivers when they drove through the extra-long tunnels and highways were extracted.The statistical methods were then applied to compare the indicators differences caused by the driving experience and road sections.The time-domain indicatorses of heart rate(HR)and heart rate variability(HRV)were taken as the characterization parameters to construct the quantitative model of driver?s mental workload using principal component analysis(PCA)and exploratory factor analysis(EFA).Then the accuracy of the model was validated by the frequency-domain analysis of HRV indicatorses.Based on the analysis of speed characteristics,this thesis proposed a risk assessment method for drivers' speed control behavior combined with the subjectively expected speed and the objectively safe speed.A quantitative model of driver's comprehensive risk was subsequently set up using PCA,and the K-means clustering analysis was adopted to identify the risk attributes of drivers.The C4.5 decision tree was utilized to establish discriminant models for two types of drivers with different risk attributes.The mental workload and the corresponding speed were regarded as the input and output variables respectively.On this basis,an autonomic neural network learning method was used to establish a time-series speed prediction model considering risk motivation.Ultimately,the characteristics of the eye movement indicators(including fixation,saccade,and blink)were analyzed to supplement the content of the visual workload and to compare the results of the mental workload model further.The driver's visual interest area was divided into seven categories using dynamic clustering,and the driver's fixation proportion was calculated.The Markov chain was then applied to calculate the one-step transfer probability and smooth distribution probability in different visual interest areas to study the driver's spatial attentional distribution and attention transfer characteristics in tunnel environment.The results showed that both skilled drivers and unskilled drivers revealed a higher mental workload far away from the tunnel entrance but closer to the tunnel exit.Familiar with the road conditions could reduce the driver's mental workload to a certain extent and speed up the transition process of the driver's physiological and psychological reaction during the entrance and exit section of the extra-long tunnels.Relative to the dark adaptation,the process of light adaptation exerted a less impact on the driver's mental workload and speed.In the tunnel section,the mental workload was ordered by the entrance section,the exit section and the medium section from high to low.Compared to skilled drivers,unskilled drivers had a worse physiological and psychological reaction,speed control in transition areas between extra-long tunnels and highways(short connecting section),showing a higher visual workload and speed risk value.Skilled drivers were more accustomed to observing traffic conditions in more distant places than unskilled drivers and had better foresight.Whether driving in tunnels or highways,drivers needed to repeatedly notice certain areas to extract enough information.And when the complexity of the environment increased,or the driving experience of the driver was insufficient,the probability of repeated notice on the target area would increase.