【作者】 李勇；

【导师】 陈维荣； 苏虎；

【作者基本信息】 西南交通大学 ， 电力系统及其自动化， 2008， 硕士

【摘要】 驾驶员模型对交通安全、交通管理和通行能力等方面的研究评估都有着重要的意义。本文以构造能准确描述城市交通车辆行驶特性的驾驶员模型为最终目标,对驾驶员驾驶行为展开研究,建立并实现了完整的驾驶员模型。本文依托成都市科技攻关计划项目“面向交通安全的客运车辆驾驶仿真培训系统的研发”,将人车分离,视驾驶员作为虚拟交通环境中的具有人工生命特征的实体看待,充分考虑驾驶员的个体特性,建立了安全距离跟驰模型、换道模型、交叉口冲突点判断模型和黄灯反应模型。与现有微观交通仿真模型相比,本模型综合了驾驶员反应时间、驾驶技能、性格特性和车辆特性等因素,考虑了车辆在不同路段的不同行驶特征,能更加真实、全面的描述实际交通车辆运行过程。其中,研究了车辆的启动加速跟驰和减速停车过程,使跟驰模型能准确描述车辆行驶的全部过程;采用了速度判断和概率判断相结合的换道意图产生模型,利用曲线拟合确定车辆换道路径;深入分析了交叉口车辆行驶规律建立了可靠的冲突点判断模型;提出了较为实用的黄灯反应模型。本文所建驾驶员模型可仿真不同风格的驾驶行为。由于虚拟个体具有不同驾驶风格和不同行为特征,因此该模型能一定程度上体现出车辆在微观层面上的多样性。本文采用现有实测数据对跟驰模型三个过程分别进行了验证,验证结果表明所建跟驰模型一定程度上能有效、准确的描述车辆的稳定跟车、启动加速跟车和减速停车跟车过程。本文考虑到了实际交通中的主要因素,利用所建驾驶员模型能比较完整的描述车辆的基本行驶过程和特征。更多还原

【Abstract】 Driver models play the role of great significance in study of traffic safety, traffic management and capability of transport system. In order to describe the character of driving behavior in city traffic, this thesis studies the behavior of driver, creates and implement an integrated driver model.This study work is based on the Chengdou Science and Technology Research Subject—"Studying in passenger transport driving simulation system for traffic safety". In the thesis, vehicle model and driver model are separated. Drivers are treated as intellective objects in virtual traffic circumstance. Individual character of drivers is considered adequately. An integrative driver model is created which is composed of car following model, lane changing model, cross conflict judging model and yellow signal lamp reaction model. Compared with modern traffic simulation model, this driver model synthesizes the reaction time, driving skill, impulse of driver and vehicle characters firstly. Considering the different running character of vehicle in different way, this driver model has capability to describe the running process of vehicle more truly and roundly. The accelerating process and decelerating process are studied particularly, so the car following model can be more precise in describing the car following process. The lane changing desire producing model is created by combining speed judgment and probability judgment, and the path of lane changing is produced by curve simulation. The running rule in cross is analyzed to found reliable conflict judge model. A feasible yellow signal lamp reaction model is founded. This driver model can simulate different driving behavior. Because the individual have different driving behavior style, this model can simulate the variety of traffic in microcosmic.The car following model is validated by data from real traffic, results show car following model founded in this thesis can availably and exactly describe the steady following, startup and shutting down process. The mainly factor of city traffic is considered, so the driver model can roundly describe the characters and running process of vehicle.更多还原