驾驶人注意分散状态建模与交通仿真研究

发布时间：2018-05-05 20:48

本文选题：交通仿真 + 驾驶人模型　；参考：《长安大学》2017年硕士论文

【摘要】：作为“人-车-路”复杂系统的重要参与者,驾驶人的感知、判断、操纵都会对其驾驶状态产生影响,造成交通拥堵乃至诱发交通事故,驾驶人因素对于交通运行至关重要。以往交通仿真模型对于驾驶人心理状态差异缺乏表现,论文分别基于STCA元胞自动机与Multi-Agent两种交通仿真方法,以驾驶人注意分散为例探索心理状态变化情况下的驾驶人模型构建方法,并对一定比例驾驶人出现注意分散时的交通流运行状态进行仿真。为研究车内次任务条件下驾驶人状态变化及其对交通流造成的影响,首先,论文基于ACT-R认知结构与Distract-R平台对驾驶人注意分散状态进行了认知模拟,获得了执行4类不同次任务时的时间消耗与注意分散比,并以该数据作为驾驶人注意分散状态库,接着,在STCA元胞自动机交通流模型的基础上,建立了考虑车内次任务影响的交通流模型,该模型修正了原有的元胞自动机减速规则,且能够通过调用驾驶人注意分散状态库获取部分模型参数,并在Matlab中进行交通流仿真实验,模拟0、10%、20%比例的驾驶人在行驶中执行车内次任务时的交通流状况,次任务类型在上述4种类型中随机抽取;另外,论文基于Multi-Agent仿真方法,依次建立路段Agent、车辆Agent及信号灯Agent,并为各个Agent设计运行和交互规则,利用NetLogo平台编程建立Multi-Agent考虑驾驶人注意分散状态的交通仿真系统并进行仿真实验,从而实现不同实验结果相互验证及补充。实验数据表明,车内次任务会对交通流造成明显影响:基于元胞自动机交通流仿真实验的结果数据显示,当10%、20%的驾驶人执行次任务时,最大流量降低21.4%、36.2%,最大车速约降低11.1%、22.2%,同时,随着执行次任务驾驶人比例的增加,流量和车速峰值所对应的密度逐渐减小;基于Multi-Agent编程方法在路段模型仿真实验中的结果显示当10%、20%的驾驶人执行次任务时,最大流量降低20.3%、37.6%,最大车速约降低15.6%、29.3%,在路网模型仿真实验中,车辆集聚峰值变大,并且消散趋势变缓,同时交叉口处车辆排队现象加剧。上述实验结果表明:两种仿真方法获得的结果基本一致,符合以往文献中的经典趋势,能在一定程度上反映驾驶人注意分散状态下交通流的变化情况。通过实验同时发现,当一定比例驾驶人处于注意分散状态时,路段内交通流整体状态会受到显著影响,出现实际通行能力降低、饱和交通密度降低等情况。
[Abstract]:As an important participant in the complex system of "man-vehicle-road", the perception, judgment and manipulation of the driver will have an impact on the driving state, causing traffic congestion and even causing traffic accidents. The driver factor is very important to the traffic operation. The previous traffic simulation model is lack of performance to the difference of the driver's psychological state. This paper is based on STCA cellular automata and Multi-Agent traffic simulation methods respectively. This paper takes the driver's attention dispersion as an example to explore the construction method of the driver's model under the condition of the change of the psychological state, and simulates the traffic flow running state of a certain proportion of the drivers when the attention is dispersed. In order to study the change of driver's state and its influence on traffic flow under the condition of in-vehicle secondary task, firstly, based on the cognitive structure of ACT-R and Distract-R platform, the cognitive simulation of the driver's attention dispersive state is carried out. The time consumption and attention dispersion ratio of four different tasks are obtained, and the data is used as the driver's attention dispersion state library. Then, based on the STCA cellular automata traffic flow model, A traffic flow model considering the effect of in-vehicle secondary tasks is established. The model modifies the original cellular automata deceleration rules and can obtain some parameters of the model by calling the driver's attention decentralized state library. The traffic flow simulation experiment is carried out in Matlab to simulate the traffic flow situation of 20% of the drivers in the vehicle, and the secondary task types are randomly selected from the above four types. In addition, the paper is based on the Multi-Agent simulation method. In turn, the agents of road sections, vehicle Agent and signal lights are set up, and the operation and interaction rules are designed for each Agent. The traffic simulation system of Multi-Agent considering the decentralized state of driver is established by using NetLogo platform, and the simulation experiment is carried out. Thus, different experimental results can be verified and supplemented. Experimental data show that the secondary tasks in the vehicle have a significant impact on the traffic flow: the results of the traffic flow simulation experiment based on cellular automata show that, when 10 or 20 percent of the drivers perform secondary tasks, the maximum flow rate is reduced by 21.40.36.2and the maximum speed is reduced by about 11.1and 22.2. at the same time, With the increase of the proportion of drivers performing sub-tasks, the density corresponding to the peak value of flow and speed decreases gradually. The results of simulation experiments on road model based on Multi-Agent programming method show that when 10% of the drivers perform sub-tasks, 20% of the drivers perform sub-tasks. In the simulation experiment of road network model, the peak value of vehicle agglomeration becomes larger, the trend of dissipation becomes slow, and the phenomenon of vehicle queuing at intersection becomes more serious, while the maximum flow is reduced by 20.3and 37.6and the maximum speed is reduced by 15.6and 29.3. in the simulation experiment of road network model, the peak value of vehicle agglomeration becomes larger, and the trend of dissipation becomes slower. The experimental results show that the results obtained by the two simulation methods are basically consistent with the classical trend in previous literature and can reflect the traffic flow changes in the condition of the driver's attention to a certain extent. At the same time, it is found that when a certain proportion of drivers are in the state of attention dispersion, the overall state of the traffic flow in the road section will be significantly affected, and the actual traffic capacity will be reduced, and the saturated traffic density will be reduced.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.254

【参考文献】