返回
类型 基础研究 预答辩日期 2017-12-03
开始(开题)日期 2014-12-12 论文结束日期 2017-10-19
地点 交通学院333会议室 论文选题来源 国家自然科学基金项目     论文字数 10 (万字)
题目 信号交叉口驾驶行为交通安全风险分析
主题词 信号交叉口,驾驶倾向,换道行为,两难选择行为,事故风险
摘要 信号交叉口交通安全是道路交通安全较为薄弱的环节之一。随着我国驾驶人数量的迅速增长,驾驶人群体间差异的增大,信号交叉口交通安全面临着新的挑战,这一挑战主要来自于驾驶人在交叉口进口道换道行为及两难区选择行为的多样性而引发的交通安全的风险。基于此,本文通过在真实道路、动态交通环境下采集驾驶人车辆行驶轨迹数据,对不同驾驶倾向的驾驶人在交叉口进口道换道行为及两难区选择行为进行定量化的研究分析,分别建立了换道行为冲突风险预测模型及两难区碰撞风险预测模型,用以分析不同驾驶倾向的驾驶行为在信号交叉口进口道的交通安全风险。本文的研究对于深入理解驾驶人驾驶倾向与信号交叉口交通安全的关系、提升信号交叉口交通安全管理水平具有重要意义。 首先,对驾驶人的驾驶倾向进行了概述,设计了用于采集驾驶倾向各种特征指标的调查问卷、模拟测试与实车测试。其中调查问卷主要获取驾驶人驾驶技能、安全意识与驾驶倾向等对驾驶行为产生影响的三个因素;模拟测试采集了驾驶人的反应时间能力与速度估计能力;实车测试采集了实验者的换道频率、驱动频率与制动频率。在此基础上,使用K-均值聚类分析法,将实验者的驾驶倾向分为:激进型、稳健型和保守型三类,并对其驾驶行为特性进行了分析。 其次,对传统的车辆行驶轨迹采集方法(高点录相法、实验车法、模拟驾驶法)的特点进行了分析。针对驾驶人换道行为,对传统的车辆轨迹采集方法进行了改进,设计了一种高点录相法与实验车法相结合的换道车辆行驶轨迹数据采集方法;针对驾驶人两难区选择行为,利用传统的录相法获取两难区的车辆行驶轨迹数据。使用Tracker软件对行驶轨迹数据进行提取,并利用相机成像理论对数据进行转换,对转换后的真实的轨迹数据进行平滑处理,进而分别计算出不同驾驶倾向驾驶人的换道行为与两难区选择行为的特征参数。 再次,从宏观与微观两个层面对驾驶行为的涵义进行了梳理与诠释,并从微观层面确定了表征驾驶行为的驾驶特性。对于换道行为,主要从换道开始执行的位置、换道持续时间、换道过程中的行驶速度以及换道过程中车辆间距等特性进行表征;对于两难区选择行为,主要从绿灯开始闪烁时车辆至停车线距离、行驶速度、加速度、减速度、车头时距等特性进行表征。在此基础上,分别对激进型、稳健型和保守型驾驶人的换道行为及在两难区的选择行为的行驶特性进行了描述与分析。 然后,以衡量冲突风险概率的碰撞时间和以衡量冲突风险严重程度的避撞减速度构建冲突风险综合指标,利用该指标确定四种风险水平:零风险水平、低风险水平、中风险水平和高风险水平。基于此,构建换道行为冲突风险预测模型,模型结果表明:纵向速度、与目标车道后车间距以及驾驶倾向这三个因素对车辆换道过程引发的冲突风险有显著性影响。与零风险水平相比,当纵向速度增加、与目标车道后车间距减小、驾驶倾向激进程度增高时,车辆换道引发冲突处于低风险水平、中风险水平和高风险水平的发生比均有不同程度的增加。模型还可以用于预测特定条件下换道行为处于不同冲突风险水平的概率,可以为换道仿真模型提供关于交通冲突的计算方法与理论依据。 最后,基于驾驶人两难区选择行为的分析结果,建立了相位过渡期间包括驾驶倾向因素在内的驾驶人二元选择(继续行驶或停车)模型,以此确定不同驾驶倾向条件下两难区的上、下边界。在此基础上,建立了追尾碰撞风险预测模型与侧向碰撞风险预测模型,使用蒙特卡罗进行模拟,结果表明:处于两难区车辆的侧向碰撞风险概率高于追尾碰撞概率;另外,即使车辆处在至停车线相同的距离,其碰撞风险也会因驾驶人的驾驶倾向不同而不同:对碰撞风险的影响程度由大到小的驾驶倾向依次为:激进型、保守型、稳健型。
英文题目 RISK ANALYSIS OF TRAFFIC SAFETY BASED ON DRIVING BEHAVIOR AT SIGNALIZED INTERSECTION
英文主题词 signalized intersection,driving tendency,lane change,driver behavior in dilemma zone, accident risk
英文摘要 Signalized intersection traffic safety is one of the relatively weak link in road traffic safety. With the rapid growth of the number of drivers in our country, the difference between the drivers is increasing, and the traffic safety at the signalized intersection is facing new challenges. This challenge is mainly due to the choice of the driver’s entrance roadway and the dilemma at the intersection The diversity of behavior and the expansion of the traffic safety assessment of the contents of the signal. Based on this, this paper chooses the driver’s vehicle trajectory data in the real road and dynamic traffic environment, and carries on the quantitative analysis to the driver of the different driving tendency at the intersection entrance road entrance behavior and the dilemma selection behavior, The conflict risk prediction model and the collision risk prediction model of the dilemma collision, reveal the influence of the different driving characteristics of the driving direction of the driver at different intersections on the traffic safety of the signal intersection. The research of this paper is of great significance to understand the relationship between driver’s driving tendency and traffic safety at signalized intersection, and improve the traffic safety management level of signalized intersection. Firstly, the driving tendency of the driver is summarized, and the questionnaire, static test and dynamic test are designed to collect various characteristic indexes of driving tendency. Among them, the questionnaire mainly obtains three factors that affect driver’s driving skills, safety awareness and driving tendency. The static test collects the driver’s response time and velocity estimation ability. The dynamic test gathers the experimenter’s lane Frequency, drive frequency and braking frequency. On the basis of this, the driving tendency of the experimenter is divided into four types: radical, robust and conservative, and the driving behavior characteristics are analyzed by using K-means clustering analysis. Secondly, the characteristics of the traditional vehicle trajectory acquisition method (high point recording method, experimental vehicle method and simulated driving method) are analyzed. Aiming at the driver ’s lane changing behavior, the traditional method of vehicle trajectory acquisition is improved. A high - point recording method and experimental vehicle method are combined to study the trajectory acquisition method. In view of the driver’ s dilemma, the recording method to obtain the dilemma of the vehicle trajectory data. Using the Tracker software to extract the trajectory data, and use the camera imaging theory to convert the data, the conversion of the real trajectory data after the smooth processing, and then calculate the different driving tendency of the driver’s lane behavior characteristics and dilemma selection Behavior characteristic parameter. Thirdly, the meaning of driving behavior is sorted and interpreted from both macro and micro levels, and the driving characteristics of driving behavior are determined from the micro level. For the track behavior, mainly from the beginning of the implementation of the track, the duration of the road, the road speed in the process of changing the road speed and the process of vehicle spacing and other characteristics of the characterization; for the dilemma selection behavior, mainly from the green light began to flash Vehicle to the parking line distance, speed, acceleration, deceleration, head time and other characteristics of the characterization. On this basis, the driving behavior of the radical, stable and conservative drivers and the driving characteristics of the selection behavior in the dilemma are described and analyzed respectively. Then, the risk prediction model of the road behavior conflict is constructed, and the collision risk is measured by the collision time of the probability of the collision risk and the collision avoidance deceleration to measure the severity of the conflict risk. Four risk levels are determined using this index: zero risk level, Low risk levels, intermediate risk levels and high risk levels. The model results show that the vertical velocity, the distance between the vehicle distance and the driving distance of the target lane have a significant effect on the conflict risk caused by the vehicle lane change process. Compared with the zero risk level, the probability of the risk level and the high risk level is different when the longitudinal velocity increases and the vehicle distance decreases from the target lane and the driving tendency is increased, the conflict caused by the vehicle lane is at a low risk level Increased degree. The model can also be used to predict the probability that the track behavior is at different risk levels under certain conditions. It can provide the calculation method and theoretical basis for the traffic simulation model for the road simulation model. Finally, based on the analysis results of the driver’s dilemma selection behavior, the model of the driver’s binary selection (continued driving or parking) including the driving tendency factors in the phase transition period is established to determine the difference between the two driving conditions, The lower boundary. On the basis of this, the risk prediction model of rear collision and the risk model of lateral collision risk are established. Monte Carlo simulation is used to show that the risk probability of lateral collision in the dilemma is higher than that of the rear collision risk. The vehicle is at the same distance to the parking line, the risk of collision will be different due to the driver’s driving tendency: the impact of the risk of collision from large to small driving in order: aggressive, conservative, steady type.
学术讨论
主办单位时间地点报告人报告主题
东南大学 2017.05.23 交通学院3楼会议室 李英帅 交叉口进口道两难区碰撞风险预测模型
东南大学 2014.4.16 东南大学中山院 刘攀教授 交通安全科研发展的历史与展望
东南大学 2014.11.28 东南院102室 徐铖铖博士 研究、撰写和发表高质量SCI论文的方法和经验
东南大学 2015.11.11 交通学院3楼会议室 杨超教授 基于手机数据的个体活动特性研究
东南大学 2015.12.11 交通学院3楼会议室 曲晓波博士 On the fundamental diagram for freeway traffic
东南大学 2016.4.29 交通学院3楼会议室 张学孔教授 智慧出行绿色交通
东南大学 2014.12.11 交通学院3楼会议室 李英帅 基于驾驶行为的信号交叉口交通安全状态评价方法
东南大学 2015.12.22 交通学院3楼会议室 李英帅 伦斯勒理工学院访学汇报
东南大学 2016.9.10 302 工作室 李英帅 驾驶倾向特征采集与聚类分析
     
学术会议
会议名称时间地点本人报告本人报告题目
香港控制工程与信息科学研究协会 2014.07 昆明 Research on integration mode of land use and traffic management
Rensselaer Polytechnic Institute 2015.10 Troy, NY,USA Development of Novice Drivers Adjustment for Capacity of Basic Roadway Segment
     
代表作
论文名称
Crash Risk Prediction Model of Lane-Change Behavior on Approaching Intersections
车辆换道行为建模的回顾与展望
Research on Integration Mode of Land Use and Traffic Management
 
答辩委员会组成信息
姓名职称导师类别工作单位是否主席备注
项乔君 正高 教授 博导 东南大学
张卫华 正高 教授 博导 合肥工业大学
郑长江 正高 教授 博导 河海大学
陈淑燕 正高 教授 博导 东南大学
季彦婕 副高 副教授 博导 东南大学
      
答辩秘书信息
姓名职称工作单位备注
梁衡弘 其他 讲师 东南大学