Research On Rear-end Collision Risk Based On Microsopic Vehicle Trajectory Data

There are millions of people who die from road traffic accidents and tens of millions of people are injured every year.The road traffic accidents have become the eighth leading cause of death in humans.Among all types of the road traffic accidents,the rear-end collision accounts for the largest proportion.Therefore,it is necessary to carry out analysis and control research on rear-end collision risks.On the basis of introducing and summarizing the research status of rear-end collision risks,this study focused on the research of the transition duration and some related critical durations of rear-end collision risk,and then explored the advantages of the connected and automated vehicle(CAV)in the prevention and control of rear-end collision risk by citing the Next Generation Simulation Community(NGSIM)database.The main research results of this study include the following points:(1)Research on the transition duration of rear-end collision risk.The rear-end collision risk does not appear instantaneously,so it is necessary to conduct research on the transition duration of rear-end collision risk.Taking the time to collision(TTC)as the safety evaluation index and considering the speed changes of car-following pairs,the direct cause of rear-end collision risk was explored from the perspectives of theoretical analysis and actual data verification.Results indicate that there are 13 types of risk causes in theory,while only 3 types of causes have a higher possibility of causing a rear-end collision risk in fact.In order to further reveal the potential causes of risks,the transition duration was defined.Besides,a new indicator,the derivative of TTC(TTCD),was proposed to describe the changes of the TTC during the transition duration;and the influencing factors of TTCD at important points were further explored based on the regression model.The changes of TTC and TTCD under the conditions of the three high-proportion causes leading to risk were compared.And then the sensitivity analysis was carried out to verify the robustness of the results.Finally,some prevention measures of rear-end collision risk were proposed based on the obtained results,which is of great value for the reduction of rear-end collision risk and road traffic accidents.(2)Research on the critical durations of rear-end collision risk.The rear-end collision risk does not appear instantaneously,nor will it disappear instantaneously.Therefore,it is necessary to conduct research on the critical durations related to rear-end collision risk.Using TTC as safety assessment index,4 key points(the TTC upper limit point,the risk occurrence point,the TTC minimum point(the highest risk point),and the risk departure point)were determined,and then 3 critical durations of rearend collision risk were determined.Subsequently,the survival analysis was applied to explore the important influencing factors of the critical durations,and to reveal the impact of drivers’ heterogeneity at the same time.Finally,the shortcomings of TTC and its related indicators in the risk evaluation were analyzed,which are the time exposed time-to-collision(TET)and the time integrated time-to-collision(TIT).Thus the superiority of analyzing the rear-end collision risk based on durations was testified.(3)Explore the advantages of CAV in the prevention and control of rear-end collision risk.CAV is regarded as an important way to improve traffic safety.However,there is a lack of real CAV tests at this stage.The existed simulation analysis lacks an accurate description of the real traffic scenes.Thus the analysis results cannot reflect the actual safety status of CAV.In this study,the CAV trajectory data was generated based on the empirical trajectory data of human drivers in risky car-following scenarios,and the safety assessment was carried out by using TTC indicator.And then we compared the safety conditions between the CAV and the human-driven vehicle.It was found that CAV can help reduce the rear-end collision risk,while there was still some risk situations.Therefore,the CAV control model parameters were continuously optimized to improve the safety performance,which would provide theoretical support for the prevention of rear-end collision risk and promote the development of road traffic safety in the future.Besides,the performance of CAV model was verified through some examples,which would help provide a reference for the design of CAV parameters.