高速公路交通流的分形特征实证分析

发布时间：2018-01-04 12:10

本文关键词：高速公路交通流的分形特征实证分析　出处：《四川大学学报(自然科学版)》2015年04期 　论文类型：期刊论文

【摘要】：高速公路交通流演化分形特征的研究有助于深刻理解高速公路交通系统的内在演化规律,为高速公路交通流的预测和控制提供理论依据.本文利用统计学方法和复杂网络可视化技术对高速公路交通流的分形特征进行实证分析.首先,利用重标极差法计算了交通流时间序列的Hurst指数和V统计量,发现不同时间标度下高速公路交通流时间序列的Hurst指数都大于0.5,并且V统计量曲线有上升趋势,说明高速公路交通流时间序列具有自相似性和长程相关性;然后,根据可视算法,将高速公路交通流时间序列转化为复杂网络,计算网络的拓扑参数,发现网络的度分布均呈幂律分布,表明该网络为无标度网络,进一步揭示高速公路交通流时间序列为分形序列.同时发现网络的平均路径长度随网络规模的增大呈对数增长,说明网络具有小世界特征.实证分析的结果对高速公路交通流量预测中时间标度的选择和预测长度的确定有重要的参考价值.本文的研究可以为揭示高速公路交通流演化的复杂性提供新的思路和方法.
[Abstract]:The study of fractal characteristics of expressway traffic flow evolution is helpful to understand the inherent evolution law of expressway traffic system. This paper uses statistical method and complex network visualization technology to analyze the fractal characteristics of expressway traffic flow. First of all. The Hurst exponents and V statistics of traffic flow time series are calculated by using the method of rescaling range. It is found that the Hurst exponents of expressway traffic flow time series are all greater than 0. 5 under different time scales. Moreover, the curve of V statistic is on the rise, which indicates that the time series of freeway traffic flow have self-similarity and long range correlation. Then, according to the visual algorithm, the time series of freeway traffic flow is transformed into a complex network, and the topological parameters of the network are calculated. It is found that the degree distribution of the network is power-law distribution, indicating that the network is scale-free. It is further revealed that the time series of freeway traffic flow is fractal, and it is found that the average path length of the network increases logarithmically with the increase of the network size. The result of empirical analysis has important reference value for the choice of time scale and the determination of forecast length in expressway traffic flow forecasting. The research in this paper can be used to reveal the highway intersection. The complexity of current flow evolution provides new ideas and methods.
【作者单位】：四川大学计算机学院;四川智能交通系统管理有限责任公司;
【基金】：国家自然科学基金(11102124,91338107) 教育部博士点基金(20130181110095)
【分类号】：U491.112;O157.5
【正文快照】： 1引言对于纷繁复杂的现象,能够全面深刻地认识其特征一直是人们不懈努力的一个方向.分形理论、复杂网络、分子动力学等理论的出现,为探索复杂性问题提供了新的思路.当前人们对复杂系统已经进行了极为广泛与深入的探索,研究涉及了商业交易[1],物流运输[2],网站浏览[3],电影点播

【参考文献】