A Study Of Overtaking Maneuvers On Mountainous Two-lane Roads Based On Trajectory Data

With the rapid expansion of China’s highway construction,along with increased network connectivity and the growing demand for urban residents’ travel,there has been a significant rise in passenger and cargo transportation on two-lane highways and smaller roads.However,this surge in traffic has also led to a higher number of traffic accidents,particularly head-on collisions resulting from improper overtaking,which are the most devastating.Addressing the issue of reducing overtaking accidents on mountainous twolane roads has become an urgent challenge for traffic management authorities,aiming to enhance safety.Relying solely on the placement of warning signs and speed limits,among other management measures,is insufficient in effectively mitigating the risks associated with overtaking accidents.Hence,it is imperative to conduct a thorough study on overtaking behavior on dual carriageway highways in order to identify the underlying risk factors in the overtaking process.By doing so,appropriate safety suggestions and measures can be formulated to improve overall traffic safety.The main purpose of this paper is to analyze and model the overtaking behavior of mountainous two-lane highways,clarify the characteristics of overtaking behavior of mountainous two-lane highways,and analyze the causes and influence mechanisms of long overtaking duration.In view of the problem of long overtaking distance on mountainous two-lane highways,the degree of satisfaction of the current specification on the actual overtaking sight distance is evaluated to provide effective support for overtaking simulation research and traffic safety improvement on two-lane highways.Firstly,we designed a data collection scheme for overtaking behavior on mountainous two-lane roads and collected relevant data,then extracted the original trajectory data based on Tracker software,analyzed the error components in the original data,and proposed a targeted trajectory cleaning and reconstruction method,i.e."adjust sampling frequency-outlier replacement-Smoothing and noise reduction".Finally,more than 30,000 trajectory data with high quality,reasonable kinematic parameters and good response to vehicle overtaking behavior were obtained,which effectively supported the analysis and modeling research of overtaking behavior of double vehicles in mountainous areas.Secondly,the statistical analysis of overtaking behavior characteristics was carried out from different perspectives of overtaking behavior composition characteristics,overtaking behavior kinematic characteristics and lateral interaction characteristics,respectively.After that,the overtaking behaviors of two-lane highways in different regions were compared and analyzed,and the characteristics of local overtaking behaviors were summarized and refined.The results show that passenger cars are the main overtaking vehicles,and passenger cars overtaking trucks is the most important overtaking behavior.The overtaking behavior of two-lane roads in mountainous areas has the characteristics of fast speed,large speed difference,overtaking distance and long duration,which may be the main reason for the frequent occurrence of overtaking accidents in mountainous areas.The lateral interaction can be mainly divided into "trapezoid" and "triangle" patterns,among which the risk of overtaking is higher in the "triangle" pattern.After that,an analysis model of factors influencing the duration of overtaking hazard exposure was constructed based on survival analysis theory,the reasons for long duration of overtaking on mountainous two-lane roads were analyzed,and the mechanisms of categorical variables and combination variables influencing the duration of overtaking hazard exposure on mountainous two-lane roads were explored.The results show that the long overtaking duration is mainly due to two types of local overtaking characteristics,the first type of overtaking behavior with large initial spacing and large speed difference,and the reluctance to return to the original lane in time after completing the overtaking,which makes the overtaking duration longer.The second type of drivers with small initial spacing and small speed difference do not accelerate significantly during the overtaking process,resulting in a long overtaking duration.Based on the Kaplan-Meier single factor effect analysis model,it was found that different overtaking vehicle types,different overtaken vehicle types,different speed difference types and the presence of oncoming traffic in the opposite direction were statistically significant for the overtaking hazard exposure duration.The modeling of hazard exposure duration based on the acceleration failure model showed that for the Weibull distribution,Lognormal distribution and Log-logistic distribution,the Log-logistic distribution fitted the best model with AIC and BIC of 272.989 and 265.650,respectively.the coefficients with the greatest degree of influence on the hazard exposure duration were variable is the type of overtaking vehicle,followed by the presence or absence of opposing vehicles has a greater effect on the duration of overtaking.Finally,the question of whether the current normative overtaking sight distance values can meet the needs of overtaking behavior in the current traffic environment is addressed for long overtaking distances on mountainous two-lane roads.The differences of the current major overtaking sight distance models are analyzed in terms of the assumptions underlying the models,the kinematic relationships of the overtaking vehicles and the speed categories selected for the calculations.The overtaking sight distance model is calibrated using the actual measurement data to determine the overtaking sight distance model with the least error in overtaking behavior on two-lane roads in mountainous areas.Based on this,a reliability theory-based overtaking sight distance assessment model for mountain two-lane highways was constructed to evaluate the failure probability of the current standard at a design speed of 60 km/h,revealing the relationship between the current specification and the actual sight distance requirements for overtaking behavior on mountain two-lane highways.The results show that the average absolute error between the Llorca model and the measured data is the smallest,only 16.67 m.When using trucks as the unfavorable overtaking vehicles for overtaking sight distance assessment on mountain two-lane roads,the reliability probabilities corresponding to the AASHTO model and the Llorca model are only 56.99% and 33.08%,which are both relatively low.This indicates that the recommended values of overtaking sight distance given in the current specification meet the actual sight distance demand for overtaking on mountainous two-lane highways to a lesser extent,and some of the recommended values may need to be reviewed.This paper analyzes and models the overtaking behavior of two-lane highways in mountainous areas by integrating theoretical knowledge from multiple disciplines such as computer science,traffic engineering and statistics.On the one hand,it enriches and expands the gap of domestic research on two-lane highway overtaking behavior based on actual measurement data.On the other hand,by proposing the overtaking sight distance assessment model based on the measured data,it helps the relevant departments to validate the recommended values of the standard specification to improve the overtaking safety of two-lane highways.