导　　师： 孙会君

学科专业： 082303

授予学位： 博士

作　　者： ;

机构地区： 北京交通大学

摘　　要： 出行者的决策行为直接影响着城市交通网络中的交通流量分布状态。对出行者的交通出行行为进行研究,是城市交通规划、交通需求管理和交通网络设计等研究的基础。本文在出行者有限理性的假设下,分别基于满意决策准则和累积前景理论对出行行为进行分析和建模,研究出行者的选择行为与交通网络流量分布的相互作用规律,探索交通需求管理策略,引导出行者的选择行为以达到缓解城市交通拥堵的目的。具体来讲,本论文的研究工作主要有以下几个方面： /(1/)研究了双模式网络中出行者有限理性的方式选择行为,建立了基于有限理性的双模式选择均衡模型,分析了最坏的、最好的和一般的三种情况下的均衡解,基于决策者不同的风险偏好/(风险追求、风险中性和风险厌恶/)研究了道路收费和公交票价策略,分别建立了优化道路收费水平的最小化最大值、最小化最小值和最小化平均值模型,运用了指南针搜索算法进行求解,通过算例分析了出行者基于有限理性的选择行为特性和决策者基于风险偏好的道路收费策略对系统总成本的影响。 /(2/)假设双模式网络中交通需求是不确定的,出行者是有限理性的,他们在选择交通方式时会综合考虑出行成本的均值和方差,并且有可能会在一个合理的阈值范围内选择感知出行成本比较高的交通方式,构建了优化公交票价的双层规划模型,上层以系统总成本均值最小为目标,下层为基于风险偏好和有限理性的方式选择均衡模型,运用了灵敏度算法进行求解,给出算例分析了需求的不确定性和出行者的有限理性对交通方式分担、公交票价和系统总成本的影响。 /(3/)在经典瓶颈模型的基础上,将“最早可接受的到达时间”、“最理想的到达时间”和“工作开始时间”作为三个参考点,同时考虑通勤者行驶时间的绝对负效用和到达时间的相对效用,建立了基于多参考点和出行效用的瓶颈道路收费模型,得到了动态收费策略,提出了两种单阶段收费策略和三种双阶段收费策略,并得出了它们各自的收费水平、收费时间和收费效果,单阶段收费策略最多能减少超过1//2的总排队时间,双阶段收费策略最多能减少超过2//3的总排队时间。 /(4/)假设出行时间服从正态分布,运用参考依赖理论研究了出行者的路径选择行为,同时考虑了出行者选择路径的内在效用和收益-损失效用,前者由路径出行时间决定,后者由路径出行时间和参考点共同决定,建立了基于参考依赖理论的随机网络用户均衡模型：假设出行时间服从自由分布,运用累积前景理论研究了出行者的路径选择行为,认为出行者总是选择具有最大的最坏情况下的累积前景值的路径,建立了基于累积前景理论的随机网络鲁棒用户均衡模型；分析了出行者的风险偏好、损失规避和参考依赖等行为特性对网络均衡结果的影响。 Travelers'decision-making behaviors in trip directly influence the distribution of the urban traffic network flow. The researches on the travel behaviors are the foundation of urban traffic planning, traffic demand management, traffic network design, and so on. Within the frame of bounded rationality, this paper intends to analyze and model travel behaviors based on satisficing decision rule and cumulative prospect theory. The interaction laws between travelers choice behavior and the distribution of the urban traffic network flow were investigated. Various traffic demand management strategies were represented to relieve traffic congestion through guiding travelers'choice behaviors. Specifically, the main contents of the dissertation are summarized as follows: /(1/) Travelers'mode choice behaviors based on bounded rationality in the two-mode network were studied. The two-mode choice equilibrium model based on bounded rationality was proposed and the equilibrium solutions under best case, worst case and average case were analyzed, respectively. Under decision makers' different risk preference /(risk seeking, risk neutral and risk averse/), the road toll and transit fare was researched. The min-max, min-min and minimize programming model for road toll optimization were proposed and solved by the compass search method, respectively. Numerical examples were presented to illustrate the influence of boundedly rational mode choice behavior and risk-based road toll strategies on the system total travel cost. /(2/) In the two-mode network, the travel demand was assumed to be uncertain and travelers were assumed to be boundedly rational. In the mode choice decision, travelers would fully consider the mean and variance of the travel cost, and might choose the travel mode with higher perceived travel cost in a reasonable threshold range. The bi-level programming for transit fare optimization was proposed, the upper-level objective was to minimize the mean of the total travel cost, whereas the lower-level problem was the mode choice equilibrium model based on risk preference and bounded rationality. Then, a heuristic algorithm based on sensitivity analysis approach was designed to solve the bi-level programming. Numerical examples were presented to illustrate the effect of demand uncertainty and bounded rationality on the modal share, transit fare and system total travel cost. /(3/) On the foundation of the classic bottleneck model, the earliest acceptance arrival time, the preferred arrival time and the work starting time were served as reference points, pricing models in a highway with bottleneck based on total travel utility were presented, commuters' travel time disutility and relative utility of arrival time were considered simultaneously, the optimal time-varying toll scheme was given accordingly. Further, two single-step toll schemes and three double-step toll schemes were proposed, and the charging level, pricing time and tolling effect of each toll scheme were obtained. The single-step toll can eliminate at most more than1//2of the total queueing time and the double-step toll can eliminate at most more than2//3of the total queueing time. /(4/) The route travel time was assumed to follow normal distribution and the reference-dependent theory was used to study travelers' route choice behaviors. The intrinsic utility and the gain-loss utility in travelers' route choice decision were fully considered, the former was determined by the route travel time and the latter was jointly estimated by the route travel time and reference point. Then, the reference-dependent theory-based user equilibrium model for stochastic network was presented. On the other hand, the route travel time was assumed to be distribution-free and the cumulative prospect theory was used to study traveler' route choice behaviors. Travelers always chose the routes with the maximum worst-case cumulative prospect value. Thus, the robust user equilibrium model under cumulative prospect theory was proposed. Finally, the influence of travelers' risk aversion, loss aversion and reference dependence on the network equilibrium result was analyzed.

关 键 词： 出行行为 有限理性 满意决策准则 累积前景理论 参考依赖理论 拥挤收费 公交票价

分 类 号： [U491]

领　　域： [交通运输工程] [交通运输工程]