Abstract:

The reliability of travel time is a key index of transportation network reliability. The heteroskedasticity occurs in the variability of travel time because of the uncertainty of traffic network. Thus the travel time cannot be accurately predicted through traditional mathematic statistic method. To cope with the characteristics of clustering, high peak and fat tail, the paper introduces ARCH model cluster in the econometrics. The models are formulated in the empirical analysis to evaluate the volatility and continuity of route travel time variability, describe the distribution of extrinsic information and analyze the response of route travel time variability to the inference of extrinsic information. Results show that ARCH model cluster adapted well to the variability of variance of travel time and corresponded to the features of traffic network. Effective evaluation of route travel time reliability is also obtained.

Key words: transportation engineering, reliability, ARCH model cluster, travel time, time series

摘要：

 行程时间可靠性是反映道路交通网络可靠性的重要指标.由于交通系统存在不 确定性，行程时间的波动具有异方差性，利用传统数理统计方法建立的行程时间预测模 型，无法对行程时间进行准确的预测.针对行程时间波动的尖峰厚尾、集群性等特点，引入 计量经济学中的ARCH 模型簇.通过实证分析，评价行程时间波动的时变性、持续性，描 述系统外部信息冲击的分布，分析行程时间波动对外部信息冲击的反应机制.结果表明， ARCH 模型簇能很好地适应行程时间序列数据的方差变化特点，与交通系统的自身特性 相吻合，能够对路径行程时间的可靠性进行有效的评价.

关键词: 交通工程, 可靠性, ARCH 模型簇, 行程时间, 时间序列