The existing dynamic traffic assignment models that deal with departure time choice generally construct utility or disutility functions based on the concepts of time-window and schedule delay. Travelers are assumed to be entirely rational and have full access to the information of dynamic traffic system. The reference points and the difference between arrival times within time-window are not considered in our assumption. There are often systematical deviations between traffic assignment results and real traffic (low. Traffic system is an uncertain dynamic system characterized as time-varying and stochastic. Constrained by the compound effect of imperfect traffic information, limited cognitive capability, value orientations and the degree of rationality, travelers are rarely able to perform their travel decisions with entire rationality. Researches of psychology and behavior science demonstrate that decisions under uncertainty appear to be bounded rational. As a descriptive theory, the Prospect Theory reveals psychological and behavioral mechanisms of people with bounded rationality, and demonstrates the patterns and human decision-making characteristics under risk. A few empirical studies experimenting on travelers have shown that travel behaviors under uncertainty, especially in the choices of departure time and route, and their attitude towards risk coincide with the main assumptions of Prospect Theory.rnFocused oncommuting trips during peak hours in the morning, this paper explores the usefulness of Prospect Theory in the dynamic traffic assignment. First, commuters are classified into early arrivals and late arrivals according to when they arrive at their workplace. The reference points for departure time and route choices are defined in a dynamic stochastic traffic network. The continuous functions of Prospect Theory for departure time and route choices in the condition of continuous traffic flow are constructed. A Stochastic Dynamic User Optimum (SDUO) model based on Random Effective Theory and its equivalent variation inequality (VI) are formulated, followed by a discussion of the model property. According to VI theorem, more than one local solution may exist. A heuristic algorithm is designed for model solution and a numerical example is presented to provide face validity of the algorithm. As is shown by the numerical example, both the dynamic departure flows between routes or OD and the satisfaction function of departure time choices have geometry similarities with the value function of arrival at the workplace. Most commuters (approximately 96. 45% ) can arrive at their workplace between the earliest acceptable arrival time and the job's starting time.rnTo sum up, this research expands the traditional departure time choice model within the frame of bounded rationality. After considering the time-varying and stochastic property of the traffic system, the established model could be used to analyze the departure time and route choices, as well as describe travelers' cognitive and psychological properties at the same time. This model and algorithm provide a theoretical basis for travel behavior analysis, dynamic route guidance and the establishment of travel demand management strategies, such as congestion pricing and staggered working hours. In addition, several directions for further research can be identified. Firstly, capacity restrictions of links and intersections and the resulted queuing delay should be taken into full consideration. Secondly, surveys and field experiments should be carried out to collect real data for model specification and estimation. Thirdly, the model validation on real traffic network is also considered as an interesting research objective.%以早高峰工作出行为研究对象,基于累积前景理论建立了一个随机动态用户最优(SDUO)交通分配模型,模型可以同时选择出发时间和出行路径,给出了等价的变分不等式,设计了求解算法并通过算例进行了验证.结果显示,路径及OD对之间的动态出发流量、出发时间选择满意函数均与到达工作地点的价值函数形态相似,绝大多数出行者(96.45％)都能在“可以接受的最早到达时刻”与“工作开始时刻”之间到达.模型拓展了传统出发时间选择模型中“时间窗”的概念和出行者完全理性假设的局限,在有限理性框架下考察出行者的决策行为,模型及算法可以为出行行为分析、动态路径诱导,以及拥挤收费和错时上下班等交通管理措施的制订提供理论依据.
展开▼
