【作者】 刘莹；

【导师】 崔洪军；

【作者基本信息】 河北工业大学 ， 交通运输工程， 2022， 硕士

【摘要】 网联自动驾驶车辆(Connected and Autonomous Vehicle,CAV)因其潜在优势获得了行业、政府和学术界前所未有的关注。随着网联自动驾驶车辆的逐步普及,道路交通流将经历网联自动驾驶车辆与手动驾驶车辆(Human-driving Vehicle,HV)混行的阶段和仅有网联自动驾驶车辆的纯自动驾驶阶段。换道决策是换道行为的基础,影响着高速公路合流区交通系统的高效运行。因此分析网联自动驾驶车辆合流区换道决策行为,构建多环境下换道决策模型,对于混有网联自动驾驶车辆的道路交通流管控有指导意义。论文主要研究工作如下:1)从道路条件、车辆条件、换道决策要素和仿真环境等方面设定合流区换道决策环境。分析目前高速公路合流区的不同形式,考虑不同类型车辆的驾驶差异性,依据Na Sch模型研究车辆合流区跟驰规则;在总结合流区换道决策模型影响因素的基础上,研究不同混入率下的网联自动驾驶车辆运行特征及决策类型;确定合流区车辆分布和到达特征以初始化交通流场景,通过MATLAB仿真软件梳理整体仿真流程和仿真参数。2)基于优化理论构建了纯自动驾驶环境下的CAV换道决策模型。分析纯自动驾驶环境的换道决策条件,基于换道决策影响因素,建立网联自动驾驶车优化模型;提出分组优化思想,并以合流区四辆车为例分析,分析分组优化的可行性,并利用遗传算法进行求解;在交通效率和环境效率两方面对优化换道决策模型进行仿真评价。结果表明,所提出的模型可以有效降低纯自动驾驶交通流中的延误,提升合流区通行效率,减少燃油消耗和尾气排放,为以全局为目标的优化决策提供指导。3)基于博弈理论构建了人机混驾环境下的CAV换道决策模型。在博弈理论的基础上分析换道行为交互,依据交互对象类型确定博弈换道类型和博弈策略,进而构建网联自动驾驶车辆支付函数和换道决策模型;在此基础上,引入协作因子λ描述网联自动驾驶车辆-网联自动驾驶车辆的协作换道特性,通过宏观交通流特征标定模型参数;以MOBIL(Minimizing Overall Braking Induced by Lane Change)模型作为基准策略,在交通效率、匀质性和安全性等方面对提出的博弈决策模型进行仿真评价。仿真结果表明,确定λ的最优取值为0.4,相较于基准模型,博弈换道决策模型下车辆速度波动次数和幅度明显减弱,碰撞时间倒数累计频率显著提高,有效提高交通稳定性和通行效率,降低碰撞风险。更多还原

【Abstract】 Connected and Autonomous Vehicle(CAV)has received unprecedented attention from industry,government and academia due to its potential advantages.With the gradual popularization of CAV,road traffic flow will go through a phase where CAVs are mixed with Human-driving Vehicles(HV)and a purely autonomous driving phase with only CAVs.The lane-changing decision is the basis of the lane-changing behavior,which affects the efficient operation of the traffic system in the expressway merging area.Therefore,analyzing the lanechanging decision of CAVs in the merging area and constructing a lane-changing decision model in multiple environments has guiding significance for the management and control of road traffic flow with CAVs.The main research work of this paper is as follows:(1)Firstly,the lane-changing decision-making environment of the merging area was analyzed from the aspects of road conditions,vehicle conditions,lane-changing decisionmaking elements and simulation environment.Through the analysis of different merging area forms in the current expressway,the appropriate form of merging area was selected to carry out the following research;considering the driving differences of different types of vehicles,the car-following rules were studied according to the Na Sch model;based on the overall consideration of the influencing factors of the lane-changing decision model in the merging area,the operating characteristics and decision types of CAVs under different mixing rates were studied.(2)Secondly,based on optimization theory,a CAV lane-changing decision model in pure CAV traffic flow environment was constructed.Through the analysis of the lanechanging decision conditions in the pure CAV traffic flow environment,an optimization model for the CAV was established based on the influencing factors of the lane-changing decision;the feasibility of grouping optimization was analyzed by taking four vehicles in the merging area as an example,and the genetic algorithm was used to solve the problem;simulation evaluation was carried out to explore the traffic efficiency and environmental efficiency of optimization model.The results show that the proposed model can reduce the delay of CAVs in purely CAV traffic flow,improve the operating efficiency of the merging area,and reduce fuel consumption and exhaust emissions.(3)Finally,a CAV lane-changing decision model in the human-machine traffic flow environment was constructed based on game theory.On the basis of game theory,the interaction of lane-changing behavior was analyzed;the type of game theory lane-changing and game strategy were determined according to the type of interaction object,and the payment function and lane-changing decision model of CAVs were constructed;on this basis,a cooperative factor was introduced to describe the cooperative lane-changing characteristics of CAVs-CAVs;the constructed model parameters were calibrated by macroscopic traffic flow characteristics;taking the Minimizing Overall Braking Induced by Lane Change(MOBIL)model as the benchmark strategy,the proposed game decision model was simulated and evaluated in terms of traffic efficiency,homogeneity and safety.The simulation results show that,the optimal value determined for is 0.4;compared with the baseline model,the number and amplitude of vehicle speed fluctuations under the game theory lane-changing decision model are significantly reduced,and the cumulative frequency of the reciprocal of Time-To-Collision(TTC)is significantly increased,which can effectively improve traffic stability,traffic efficiency and reduce collision risks.更多还原