交通诱导信息对路网中车辆行为的影响  

Impact of traffic guidance information on vehicle behavior in network

作　　者：魏赟[1] 范炳全[2] 韩印[2] 干宏程[2] 

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093 [2]上海理工大学管理学院,上海200093

出　　处：《交通运输工程学报》2009年第6期114-120,126,共8页Journal of Traffic and Transportation Engineering

基　　金：上海市重点学科建设项目(S30504);上海市科委科技人才计划项目(09QT1400400)

摘　　要：为研究诱导模型的诱导效果,用元胞自动机模型模拟车辆在路网中的行为,仿真了不同诱导信息在不同交通量、不同受诱导率情况下对交通流的影响,提出基于Agent的交通诱导模型,模型采用Q-学习算法优化诱导信息,可根据路网中交通流情况发布建议性诱导信息,调节交通流分布。仿真结果表明:影响诱导效果的主要因素为受诱导率和诱导信息,基于Agent的交通诱导模型能有效均衡路网交通流,且随着交通流的增加,优势逐渐明显。在轻交通量情况下,该模型较出行者自由选择路径模型略优;但在重交通量情况下,发布建议性的诱导信息比描述性诱导信息能减少12%平均行程时间。To study the effect of traffic guidance model, cellular automata（CA） model was applied to simulate vehicles＇ behaviors in traffic network, and the impacts of different guidance informations on traffic flow were studied in different traffic volumes and different accepting guidance ratios. A traffic guidance model based on agent technology was built. In which Q- learning algorithm was used to optimize traffic guidance information, propositional information could be provided according to real time network traffic, traffic flow distribution was adjusted. Simulation result shows that the factors that impact guidance effect in network are accepting guidance ratio and guidance information, traffic guidance model hased on agent can effectively balance network traffic flow, which is better to heavy traffic flow. In light traffic flow, the model has a little advantage compared with normal model, but in heavy traffic flow, propositional guidance information can decrease 12G average travel time compared with descriptive guidance information. 3 tabs, 6 figs, 8 refs.

关 键 词：交通诱导 智能体 微观交通仿真 元胞自动机 Q-学习算法 

分 类 号：U491[交通运输工程—交通运输规划与管理]