考虑时空分布特征的混合自行车流元胞自动机模型仿真方法  

A Cellular Automaton Simulation Model Considering Spatial-temporal Distribution for Mixed Bicycle Flows

作　　者：曹淑超 孙菲阳 李阳 CAO Shuchao;SUN Feiyang;LI Yang(School of Automotive and Traffic Engineering,Jiangsu University,Zhenjiang 212000,Jiangsu,China)

机构地区：[1]江苏大学汽车与交通工程学院,江苏镇江212000

出　　处：《交通信息与安全》2022年第2期98-107,共10页Journal of Transport Information and Safety

基　　金：国家自然科学基金项目(72001095);中国博士后科学基金项目(2020M681507)资助。

摘　　要：针对非机动车交通流中传统元胞自动机模型主观定义时空参数,粗略划分自行车虚拟车道,导致仿真精度偏差的问题,构建了精细元胞自动机模型。基于NaSch模型的更新规则,考虑二维空间内异质自行车间的错位冲突及动态换道行为特征,细化了模型网格密度及模拟时间步长。产生换道需求的自行车可以换至满足安全侧向换道条件及前行需求条件的横向位置,随后综合考虑各位置前向及侧向间隙选取最优的目标位置执行换道。在周期边界条件下,量化时空参数取值对混合自行车交通流仿真结果的影响。在镇江市正东路开展观测实验,并基于实测数据获得轨迹时空图,在宏观与微观层面验证模型的可靠性。结果表明:网格密度及时间步长取值对仿真流量影响显著,仿真流量与纵向网格密度呈正相关,与横向网格密度呈负相关,而全局网格密度的影响反映了横向与纵向网格密度的复合效应;当道路占有率为0.1时,仿真流量几乎不受时间步长的影响,而当道路占有率为0.3,0.5和0.7时,仿真流量随着时间步长的增大先增加后减小;自行车适度换道可以提高道路通行能力,但换道过于频繁则会导致交通拥堵,不同道路占有率条件下的轨迹时空特征存在明显差异,自行车流越密集,走停现象越显著;当全局网格密度为5,时间步长为0.5 s时,模拟与实测的交通流量平均绝对百分比误差为14.84%,此时拟合效果最优。Traditional cellular automata(CA)models provide inaccurate simulation results in modeling non-motorized traffic flow,due to the fact that they subjectively define spatial-temporal parameters and roughly represent bicycle lanes. With this,an improved CA model is proposed in this paper. Specifically,the grid density and time step of the proposed model are upgraded based on the updating rules of a NaSch model,which considers the conflict between heterogeneous bicycles and the dynamic lane-changing behavior in a two-dimensional space. In the proposed model,bicycles that need to make lane-changing can change to a lateral position which meets the condition of safe lateral and the forward movement. The bicycle can change lanes to the optimal position considering both the forward and lateral distance of each position. In addition,the influence of different spatial-temporal parameters on simulation results is quantified under the period boundary condition. Data from Zhengdong Road in Zhenjiang is obtained,and the spatial-temporal diagrams of trajectories are generated and with which the reliability of the proposed model is verified at both the macro and micro levels. Study results are supportive for the following conclusions.First,grid density and time step have a significant impact on the simulated flows and they are positively correlated with the longitudinal grid density but negatively correlated with the lateral grid density,and their global grid density is the compound effect of the two densities. Second,the flow is almost unaffected by the size of time step when the occupancy rate is around 0.1,but when the occupancy rates is around 0.3,0.5,or 0.7,the bicycle flow shows similar trend that first increase and then decrease with the increment of time step. Third,moderate lane-changing behavior of bicycles can improve road capacity,while frequent lane-changing behavior would lead to congestion. Significant differences in the spatial-temporal diagrams of trajectories are found under different occupancy conditions,and bic

关 键 词：非机动车交通 混合自行车流 时空分布特征 元胞自动机模型 换道行为 

分 类 号：U491[交通运输工程—交通运输规划与管理]