交通流时间序列的分形实证研究  

Empirical Study on Fractal Characteristics of Traffic Flow Time Series

作　　者：张勇[1] 关伟[1] 

机构地区：[1]北京交通大学交通运输学院,北京100044

出　　处：《公路交通科技》2010年第5期100-103,116,共5页Journal of Highway and Transportation Research and Development

基　　金：国家自然科学基金资助项目(60874078;60834001);国家高技术研究发展计划(八六三计划)资助项目(2006AA11Z212);国家重点基础研究发展计划(九七三计划)资助项目(2006CB705507);高等学校博士学科点专项科研基金资助项目(20070004020);新世纪优秀人才支持计划资助项目(NCET-08-0718)

摘　　要：为了分析交通流在同步流、拥挤流、自由流状态下的分形特征,通过关联维刻画交通流的复杂性,通过Hurst指数研究交通流的长程相关性,通过记忆长度确定交通流的影响范围。对实测数据的计算结果表明,3种交通流都是分形的,且存在长程相关性,同步流的关联维小于拥挤流,拥挤流的关联维小于自由流,同步流和拥挤流的Hurst指数大于自由流,同步流的记忆长度大于拥挤流,拥挤流的记忆长度大于自由流,说明不同状态的交通流具有不同的分形特征。In order to analyze the fractal characteristics of traffic flow under synchronization, jam and free conditions, the complexity of traffic flow was depicted by using correlation dimension, the long-term dependence of traffic flow was studied by using hurst index, and the influence area of traffic flow was delimited by using memory length. Computation result from real traffic data shows that （ 1 ） three kinds of traffic flow are fractal and exist long- term dependence; （2） correlation dimension of synchronization traffic flow is smaller than that of jam traffic flow, and correlation dimension of jam traffic flow is smaller than that of free traffic flow; （3） hurst indexes of synchronization traffic flow and jam traffic flow are larger than that of free traffic flow, memory length of synchronization traffic flow is longer than that of jam traffic flow, and memory length of jam traffic flow is longer than that of free traffic flow. It indicates that different kinds of traffic flow have different fractal characteristics.

关 键 词：交通工程 分形 时间序列分析 交通流 关联维 HURST指数 记忆长度 

分 类 号：U491[交通运输工程—交通运输规划与管理]