Traffic Equilibrium in Congested Road Networks and its applicability

本研究旨在建立拥挤路网交通流分析方法。本文尝试了多种方法，并取得了如下结果.将拥挤路段的行程时间分解为非拥挤路段的运行时间和路段末端的延误。通过求解具有明确路段容量约束的固定需求交通均衡问题，得到了均衡流量和延误.作为该问题的数值求解方法，提出了拉格朗日乘子法和内点罚函数法。前者要求严格，但在大规模网络中不实用。相反，后者是实用的，但相当近似。建立了两个考虑用户路径选择和交通信号控制相互影响的交通均衡模型。模型之一是增加交叉口交通管理能力条件。该模型的解等价于符合Smith响应控制策略Po的均衡。另一种模型要求同时求解用户均衡问题和最优信号设置问题。这个问题的均衡解对应于博弈论中的纳什均衡。提出了这两种模型的数值求解方法.利用微机建立了交通均衡分配的计算机辅助支持系统。编制了计算交通均衡的有效程序。开发了掌握路网交通流量的彩色图形显示方法和路网形态的输入方法.该系统已应用于冈山县南部的道路网。通过计算表明，模型的估算流量与实测流量吻合较好。但关于旅行时间的再现，没有获得足够的数据。这仍然是今后研究的主题。

The purpose of this study is to establish the way to analyze the traffic flow in congested road networks. Many approaches were tried and the following results were obtained.1. The travel time on congested link can be decomposed into the running time by non-congested flow and the delay at the end of link. The equlibrium flow and the delay can be obtained through the fixed demand traffic equilibrium problem with explicit link capacity restraints.2. As the numerical solution method of this problem, the Lagrange multiplier method and the interior penalty function method were proposed. The former is strict but unpractical in large scale network. On the contrary, the later is practical but rather approximate.3. Two traffic equilibrium models were formulated considering the interaction between user's route choice and traffic signal control. One of the model is to add the capability condition of traffic management at intersection. The solution of this model is equivalent to the equilibrium consistent with Smith's responsive control policy Po. The other model is to request the simultaneous solution of user's equilibrium problem and optimal signal setting problem. The equilibrium solution of this problem corresponds to the Nash equilbrium in game theory. The numerical solution method of these two models were proposed.4. The computer aided support system with respect to the equilibrium traffic assignment was made using personal computer. Efficient programs to compute traffic equilbrium were prepared. Colour graphic display methods to grasp traffic flow on road network and the way how to input the form of road network were developped.5. The above system was applied to the road network in southern part of Okayama prafecture. Through the computation, it was made clear that the estimate flows of the model fit well to the observed flows. But with respect to the reappearance of travel time, sufficient data were not obtained. That remains as the subject for a future study.