考虑有限理性的活动—出行配流模型研究

发布时间：2018-01-07 22:31

  本文关键词：考虑有限理性的活动—出行配流模型研究　出处：《南京大学》2017年博士论文　论文类型：学位论文

  更多相关文章： 活动-出行分配 准动态用户均衡 动态用户均衡 超网络 有限理性 前景理论 参照依赖理论 可接受规则 异质性

【摘要】：城市道路网络中交通流量的时空分布是出行者在参与活动的需求驱动下,在时间、城市空间、天气、交通政策等约束下,对活动内容、活动目的地、到达活动目的地的出行路径、出行方式、出发时间等内容进行选择的结果。本文基于行为决策理论对微观出行者活动-出行有限理性选择行为进行描述,基于交通分配均衡理论对宏观活动-出行流量的分布进行研究：同时,考虑到出行者之间存在的差异性是影响活动-出行时空分布多样性的重要因素,从不同角度对考虑出行者异质性的活动-出行分配问题进行了研究,主要内容如下：(1)首先,阐述出行者的活动-出行决策内容与决策过程,结合行为决策理论分析活动-出行决策过程中的有限理性行为,并基于归因理论分析影响微观个体决策以及宏观流量分布的因素；分析结果认为,有限理性主要体现在出行者对活动-出行备选方案集的有限感知、备选方案的价值评估、可能状态的主观感知以及非“最优”的决策原则四个方面,各行为决策理论则考虑了一个或几个方面的有限理性行为,比如,前景理论描述了决策者对备选方案价值评估与可能状态的主观感知两个方面的有限理性行为；另外,客观备选方案的多样性以及个体之间或者个体本身的决策差异性是影响宏观流量分布多样性的两个主要因素。(2)然后,分析了描述出行者活动-出行决策内容的多种超网络结构,并根据研究的时间范围以及离散时间间隔的长度将基于超网络的活动-出行分配理论模型进行了分类；分析表明,多状态多模式超网络具有较强的灵活性和可扩展性,适用于解决long-term的交通规划、交通需求预测等问题：而时空扩展超网络则在考虑拥堵的交通问题、short-term的交通管控问题中具有较好的应用性。(3)根据上述分析,首先对不确定性条件下的活动-出行配流问题进行了研究,回顾了描述不确定状态下出行者有限理性行为的前景理论与累积前景理论；应用简化的多状态多模式超网络表述出行者活动-出行的时空序列；假设出行时间与活动时长分别服从一定的随机分布,基于累积前景理论定义了活动-出行路径效用函数,提出了参照点离散内生的静态用户均衡模型,并通过算例分析了算法的收敛性以及模型的性质。(4)其次,对考虑出行者感知误差的准动态活动-出行配流问题进行了研究,基于参照依赖理论描述出行者参照依赖的出行路径、出发时间、活动内容以及活动目的地等问题的选择行为；假设出行者之间具有不同的参照点,提出了参照点异质的准动态随机用户均衡模型；该模型将时间离散划分为时间间隔较长的序列,假设每个时间段内,出行者完成对出行路径、出发时间、活动内容以及活动目的地的选择,并通过算例对模型的性质进行了分析；(5)再次,考虑到随着移动互联网技术的发展,出行者对各出行模式的出行时间、费用的感知近似完全信息,对动态活动-出行配流问题进行研究。基于参照依赖理论描述出行者对出行方式、出发时间、活动内容以及活动目的地等问题的选择行为；将时间离散划分为时间间隔很短的序列,假设出行者之间具有不同的参照点,提出了参照点异质内生的动态用户均衡模型,并与无参照点及参照点异质外生两种情况进行了对比；通过算例对比分析了无参照点、参照点外生与参照点内生下的流量分布状态,结果表明,参照点以及出行者对不同属性的损失规避程度对出发时间、出行方式选择的影响较大；(6)最后,考虑到不同的出行者具有不同的选择集合,对考虑决策集异质的动态活动-出行配流问题进行了研究。基于可接受规则描述出行者非“最优”决策目标的有限理性行为；假设出行者具有不同的出行方式选择集合,提出了决策集异质的可接受动态用户均衡模型；另外,新型交通出行方式的出现(如专车、共享车等)使得出行者在出行方式决策集上的异质性更为显著,将free-floating共享车作为出行者的分类依据,基于可接受规则提出了考虑共享车的动态活动-出行配流模型；讨论了各活动地点共享车供需的动态交互过程、定价、停车收费机制等问题；并通过两个算例分析了模型的性质,结果表明,共享模式的定价、活动地点的属性、停车费用等因素对共享车的选择以及活动-出行选择结果影响较大。
[Abstract]:City road network traffic flow distribution is travelers participate in the activities of demand driven, at the time, the space of the city, weather, traffic policy under the restrictions of activities, activities reached the destination, travel route, destination travel activities, departure time etc. were selected based on the results. The behavioral decision theory of micro traveler activity travel limited rational choice behavior was described, the equilibrium traffic assignment theory to study the distribution of activity travel based on traffic: at the same time, important factors to take into account the difference between the traveler is activity travel distribution diversity, has conducted the research to consider Walker heterogeneous activity travel distribution problem from different angles, the main contents are as follows: (1) first, this traveler's activity travel decision content and decision-making process, combined with the line Activity travel limited rational behavior in decision-making process for decision making analysis, and based on the attribution theory analysis of the impact of macro factors and micro individual decision-making flow distribution; the analysis results indicated that the limited rationality is mainly reflected in the traveler on the activity travel options set a limit value perception, evaluation of alternatives, the subjective perception of possible state the decision and the principle of non "optimal" four aspects of the behavioral decision theory is considered the limited rational behavior, one or several aspects such as prospect theory describes the decision value of limited rational behavior two aspects of subjective perception and evaluation of possible states of alternative price; in addition, the difference of decision objective alternative diversity as well as between individuals or individuals themselves are two main factors affecting the flow distribution of macro diversity. (2) and then describes the analysis Super network structure a variety of Walker activity travel decision content, and according to the scope of the time and the discrete time interval length will exceed the network activity travel distribution model based on the theory of classification; analysis showed that multi state multi model super network has strong flexibility and scalability, is suitable for solving the traffic long-term planning, traffic demand forecasting: temporal expansion of super network in consideration of traffic congestion, and has better application traffic control problem in short-term. (3) according to the above analysis, firstly, under conditions of uncertainty of the activity travel assignment problem is studied, reviewed the description of uncertain conditions the prospect theory of bounded rationality behavior and cumulative prospect theory; application of simplified multi state multi model super network express time series travel activities - the assumption of a walker; Time and activity time obeys certain distribution, cumulative prospect theory defines the activity travel path based on utility function, put forward reference static user equilibrium model of discrete endogenous, and through the example analysis of the convergence of the algorithm and the properties of the model. (4) second, to consider a quasi dynamic activity - Walker perceptual error travel assignment problem is studied, according to dependency theory describe the travel path, travel reference dependent departure time based on choice behavior, activities and activities to such problems; with different reference points between hypothesis travel, the quasi dynamic stochastic user equilibrium model reference point heterogeneity; the model will be divided into discrete time sequence of a time interval, assuming that each time traveler completed on route, departure time, activities and activities The choice of properties and cases of model was analyzed through calculation; (5) once again, considering with the development of mobile Internet technology, the traveler on the trip mode of travel time, the cost of the perception of near perfect information, carry out research on the dynamic activity travel assignment problem. Reference dependence theory to describe the traveler on the way to travel, the departure time based on the choice behavior of activities and activities to such problems; will be divided into discrete time series of short time interval, assuming that travelers have different reference points, put forward according to dynamic user equilibrium model of heterogeneous endogenous, and with no reference points and reference points exogenous heterogeneity was done in two cases; through the example comparative analysis on the reference point, reference points and reference points of endogenous exogenous flow distribution. The results show that the reference point and travel to different genus The loss aversion on the departure time, influence the choice of travel mode greatly; (6) finally, considering that different travelers have different sets of choices of dynamic activity travel decision in heterogeneous assignment problem is studied. The acceptable rules describe the limited rational behavior of non optimal decision target Walker based on the hypothesis; travelers have different travel mode choice set, put forward the decision set of heterogeneous acceptable dynamic user equilibrium model; in addition, the new travel mode (such as car, car sharing etc.) makes the traveler on the way to travel on the heterogeneity of the decision set is more significant, free-floating car sharing as according to the classification of walkers, acceptable rules considering dynamic activities - sharing car travel assignment model based on; discuss the venue share dynamic delivery of supply and demand interaction process Pricing, parking fee mechanism and other issues. Two examples are given to analyze the characteristics of the model. The results show that factors such as pricing of shared mode, location of activities, parking cost and other factors have great influence on the choice of shared cars and the result of activity trip selection.

【学位授予单位】：南京大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：U491
，

本文编号：1394539