International comparison on characteristics of travel behaviour in congested network

本研究旨在获得世界各地城市拥堵网络中出行者行为的经验知识。这项研究开发了一种在成本较低的城市测量拥堵程度的方法。应用沃德罗的均衡理论，它打算减少要测量的链路数量。假设预先获得网络的静态信息，例如节点、链路、链路的自由流动行进时间以及它们的连通性的拓扑信息。要测量的链路由该信息确定。使用该方法，可以根据有限路段上的行程时间的测量结果来估计未进行测量的路段的行程时间。上述方法是基于路线的理论描述的，因此有必要列举每个O-D对之间的所有可能的路线。然而，由于计算成本的原因，枚举所有路由是不现实的。这一理论提供了一种无需枚举即可完成计算的方法。另一方面，现实世界中的旅行者可能没有能力列举出他们的出发点和目的地之间的所有可能的路线。他们可能会从他们可以考虑的有限选择中选择自己的路线。考虑到出行者的这种行为，本研究提出了一种从所有可能的路线中提取出行者的路径选择选项的算法。对该方法进行了实证检验，并对世界上选定的城市进行了实证调查。网络的行程时间由配备GPS的汽车获得。得到了路网各路段的行程时间。将这一结果与上述方法进行比较，可以得到拥塞网络中的拥塞特征和出行行为。

This research intends to obtain empirical knowledge of travellers' behaviour in congested network in cities in the world. This research developed a methodology for measuring degree of congestion in cities with lower cost. Applying Wardrop's theory of equilibrium, it intends to reduce the number of links to be measured. Static information of the network, such as nodes, links, free flow travel time of links, and topological information of their connectivity, is assumed to be obtained beforehand. The links to be measured are determined by this information. Using this methodology, travel times of links where measurement is not made can be estimated from the result of measurement of travel time on limited links.The methodology explained above is described by route based theory and therefore enumerating all possible routes between each O-D pairs is necessary. However, it is known that enumerating all routes is not realistic due to calculation cost. This theory provides a method to finish calculations without such enumerations. On the other hand, travellers in the real world may not have ability to enumerate all possible routes between their origins and destinations. They may choose their routes from the limited options they can consider. Considering such behaviour of travellers, this study shows an algorithm to extract route choice options of travellers from all possible routes. This method was checked by empiricl test.The empirical surveys were made in the selected cities in the world. Travel time of the networks was obtained by GPS-equipped cars. Travel time of all links of the networks has been obtained. Comparing this result with the methods shown above, characteristics of congestion and travel behaviour in congested network can be obtained.