基于贝叶斯网络的公交停靠站时两轮车穿插行为研究  

Research on Crossing Behavior of Two-wheeler at Bus Station Based on Bayesian Network

作　　者：邵海鹏[1] 李静娴 谢沛 慕伟[1] 

机构地区：[1]长安大学公路学院,西安710064 [2]长安大学信息学院,西安710064

出　　处：《武汉理工大学学报（交通科学与工程版）》2017年第6期969-973,共5页Journal of Wuhan University of Technology（Transportation Science & Engineering）

摘　　要：城市道路中,公交车辆停靠站时往往会占用非机动车道,阻碍非机动车及摩托车的通行,为了探讨在公交车站停靠站时两轮车穿插行为,分析了在城市道路中,影响公交车辆停靠站时两轮车穿插行为的八个影响因素,以896组观察实验数据作为分析依据,基于专家知识和数据融合方法建立了公交车停靠站时两轮车穿插行为的贝叶斯网络结构,利用服从Dirichlet分布的贝叶斯方法对贝叶斯网络进行参数学习,结合网络模型,应用联合树引擎推断了在各个变量影响下两轮车穿插行为概率分布.结果表明,驾驶员的性别、驾驶非机动车的种类、左侧机动车辆的车头时距及交通干扰环境对非机动车驾驶员的穿插行为具有直接的影响.In urban roads,bus stopping often occupies non-motor vehicle lanes and hinders the passage of non-motor vehicles and motorcycles.To explore the crossing behavior of two-wheeler in the bus station,eight factors influencing crossing behavior of two-wheeler at bus station is analyzed firstly.Based on 896 groups of observed data,the Bayesian network structure of crossing behavior of twowheeler at bus station is obtained by expert knowledge and data fusion method.Then the Bayesian network parameter learning is conducted by Bayesian method subjected to Dirichlet distribution.Finally,the probability distribution of non-motorized crossing behavior under the influence of each variable is deduced by junction tree algorithm combined with the network model.The results show that driver’s gender,the type of non-motor vehicle,the instant time headway of motor vehicle on the left and the traffic environment have a direct influence on the crossing behavior of non-motor vehicle drivers.

关 键 词：交通秩序 穿插行为 公交车站点 贝叶斯网络 K2算法 

分 类 号：U491.2[交通运输工程—交通运输规划与管理]