【作者】 郑亮；

【导师】 马寿峰；

【作者基本信息】 天津大学 ， 系统工程， 2013， 博士

【摘要】 微观交通流模型在重现交通流基本特征和交通管控过程中发挥着重要的作用。因此，建立符合实际的微观交通流模型将有利于更好地管理和控制交通系统的运营状态。而以往大部分的微观驾驶行为建模主要考虑了前向交通信息刺激，忽略了后向或侧向的车辆信息对驾驶行为的刺激作用，导致了基于此类模型的仿真交通与复杂的实际交通存在一定的偏差，影响了制定合理的交通管控措施。因此，针对复杂交通场景中驾驶员的行为习惯，采用元胞自动机模型和优化速度模型分别分析了后向和侧向车辆信息对驾驶行为的作用，提出了双向刺激的元胞自动机模型和双车道优化速度模型。本文主要的研究工作如下：1.日益成熟的车间通信技术让驾驶员主动获取后向车辆信息成为可能。因此，在NaSch模型中引入双向车辆信息对驾驶行为的作用机制，提出了主动获取后向车辆信息的双向刺激的元胞自动机模型。同时，根据对危险感知的敏感程度可将驾驶员分为激进型、一般型和保守型，并通过随机慢化概率的大小来描述。最后，通过一系列的数值实验可知：驾驶行为越激进或车流状态越拥堵，车队的行驶安全性越低；考虑双向车辆信息的作用将有利于提高车队的行驶安全性；驾驶员敏感类型和信息刺激类型对异质车队的行驶安全性产生重要影响。2.采用不同的鸣笛或开大灯条件也可以描述三种类型的驾驶员：激进型、一般型和保守型，并在启发式元胞自动机模型的基础上考虑了驾驶员对鸣笛或大灯的响应机制，提出了被动接收后向车辆信息的双向刺激的元胞自动机模型。最后，通过数值实验的方法可得：驾驶员对被动接收的鸣笛或大灯刺激的响应将改进车辆的动力行为，但是却降低了匀质/异质车队的行驶安全性。3.在横向摩擦存在的前提下，分析双车道上车辆的动力行为过程，提出了带横向摩擦的双车道优化速度模型并采用H∞范数的方法获得了基于此模型的交通流的稳定条件。然后，加载了两种不同的信息刺激：无时间延迟的相对速度信息和有时间延迟的车头间距信息，同样运用H∞范数的方法求得基于此改进模型的交通流的稳定性条件。最后，通过数值实验的方法验证了所得的理论结果，并演示了换道行为对交通流稳定性的负面影响。4.考虑到驾驶员的有限理性，采用视角信息替代相应的3-D交通信息，并考虑不同车道车辆之间的横向摩擦程度，提出了考虑横向摩擦程度的双车道优化速度模型并运用线性稳定理论获得了基于此模型的交通流的稳定条件。最后，数值实验验证了：减少横向摩擦程度、增强驾驶员的敏感系数或较大的车辆尺寸都将有利于交通流的稳定性，并演示了即使发生分离、合并或换道行为，减少横向摩擦都将有利于车流的稳定性。更多还原

【Abstract】 Microscopic traffic flow models play an important role in reproducing thefundamental characteristics of traffic flow、managing and controlling the traffic flow.Therefore, establishment of realistic microscopic traffic flow models will beconductive to better manage and control the traffic systems. However, most previousmicroscopic modeling about driving behaviors mainly considers the stimuli from theforward traffic and neglects the stimulating effect from backward or lateral vehicles,which causes the deviation between the simulated traffic based on this class of modelsand complex real traffic and finally affects designing the rational measures for trafficmanagement and control. Therefore, due to the driving behaviors in complex trafficscenarios, this paper employs the cellular automaton model (CA) and optimal velocity(OV) model to respectively study the impact of backward and lateral vehicleinformation on the driving behaviors, and then proposes the bi-directional stimulatedcellular automaton model and two-lane optimal velocity model. The main researchwork can be summarized as follows：1. Development of vehicle to vehicle communication technology makes itpossible for drivers to actively obtain the backward vehicle information; therefore, theinfluence of bi-directional vehicle information on driving behaviors is involved inNaSch model. Meanwhile, the drivers can be classified to three types (i.e. aggressive、normal and conservative) based on their sensitivity to the danger, which can bedescribed by the possibility of random braking. Finally, through numericalexperiments it is known that more aggressive driving behaviors or more congestedtraffic causes lower safety performance; while the bi-directional vehicle informationcontributes to increase the safety performance; the sensitive and stimulated types ofdrivers produce a great effect in the safety performance of the heterogeneous fleets.2. Applying the various conditions of activating the honk or headlamp can alsodescribe three types of drivers, and the mechanism of drivers response to the stimuliof honk or headlamp is taken into account based on the heuristic CA model, based onwhich the bi-directional stimulated CA model involving the backward vehicleinformation received passively by drivers is proposed. Finally, it is obtained throughnumerical experiments that drivers response to the honk or headlamp stimulusimproves the dynamic behaviors of vehicles, but reduces the safety performance ofhomogeneous/heterogeneous fleets. 3. Under the prerequisite of lateral friction, the dynamic behaviors of vehicles intwo-lane road are analyzed, and then a two-lane OV model with lateral friction is putforward, based on which the stability condition is solved by the method of H∞norm.After that, two kinds of various information stimuli are added to the proposed model,which are relative velocity information without time delay and headway distanceinformation with time delay, and their stability conditions can also be gained by themethod of H∞norm. Finally, numerical results verify the obtained theoreticalconclusions and demonstrate the negative impact of lane changing behavior on thetraffic flow stability.4. Because of the drivers bounded rationality, the visual angle information isemployed to substitute the corresponding3-D traffic information, based on which thetwo-lane OV model with lateral friction between vehicles is presented. After that, thestability conditions of traffic flow based on the proposed model are obtained by thelinearized stability theory. Finally, numerical results confirm that decreasing thelateral friction, increasing the sensitivity of drivers and larger vehicle size all benefitthe stability of traffic flow, and also illustrate that reducing the lateral friction iscontributive to the traffic flow stability even though the diverging、emerging or lanechanging behaviors happen.更多还原