交通流时间序列的分形实证研究

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。

Empirical Study on Fractal Characteristics of Traffic Flow Time Series

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 longterm dependence;(2)correlation dimension of synchronization traffic flow is smaller than that of jam traffic flow,and correlation dimension of iam 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.