公交出行中的个体学习行为及其仿真的研究

发布时间：2018-08-26 11:33

【摘要】：以公交系统中的个体学习行为为研究对象,通过构建出行者面对动态出行环境的选择行为模型,使用基于多智能体的仿真方法,模拟公交运营服务过程以及出行者的学习与自适应行为,捕捉个体行为所涌现的宏观特性。通过对个体的心理认知特性及其公交出行过程进行描述和分析,建立了个体出行行为的理论模型和方法框架。论文通过阐述马尔科夫决策过程的基本属性,解析公交服务的具体过程,建立了个体公交出行的路径选择模型,并给出了模型的强化学习算法;通过对个体公交出行感知更新机制和出发时刻选择过程的分析,建立了个体公交出行的感知更新与出发时刻选择模型。通过仿真实验发现:上述理论模型和方法能够较好的表征个体在公交出行中的学习行为和自适应过程。分析了交通信息在公交出行者感知学习过程中的作用。提出了一个基于Q学习参数的信息评价机制和信息信任指数更新机制。通过仿真实验发现:交通信息在出行中的主要角色是外部影响要素,它会提高出行者对环境变动的敏感程度,但是并不能给出行者带来额外的收益;只有当环境发生突变时,交通信息才能发挥显著的作用。分析了个体在公交出行中的习惯选择行为。通过分析习惯选择与主动学习的关系,提出了一个考虑用户习惯选择行为的个体公交出行模型以及习惯选择的触发和终止机制。通过仿真实验发现:在不确定性的出行环境中,多数出行者的习惯选择行为是系统走向均衡的有益因素。最后,在Swarm平台的基础上,设计并开发了基于多智能体的微观仿真系统,实现了针对上述理论模型的微观仿真实验。
[Abstract]:Taking the individual learning behavior in the public transport system as the research object, the model of travelers' choice behavior in the dynamic travel environment is constructed, and the simulation method based on multi-agent is used. The process of bus operation and service and the learning and adaptive behavior of travelers are simulated to capture the macroscopic characteristics of individual behavior. Based on the description and analysis of individual psychological cognitive characteristics and public transportation travel process, the theoretical model and method framework of individual travel behavior are established. Through expounding the basic attribute of Markov decision process, analyzing the concrete process of public transport service, establishing the path choice model of individual public transport trip, and giving the reinforcement learning algorithm of the model; Based on the analysis of the mechanism of individual bus trip perception renewal and the process of departure time selection, the model of individual bus trip perception renewal and departure time selection is established. The simulation results show that the above theoretical models and methods can well represent the individual learning behavior and adaptive process in public transportation. This paper analyzes the role of traffic information in the process of public transport travelers' perceptual learning. An information evaluation mechanism based on Q learning parameters and an update mechanism of information trust index are proposed. The simulation results show that the main role of traffic information in travel is external factors, which will enhance the sensitivity of travelers to environmental changes, but can not bring extra benefits to travelers; only when the environment changes, Traffic information can play a significant role. The habit choice behavior of individual in public transportation is analyzed. Based on the analysis of the relationship between habit selection and active learning, an individual bus travel model considering the user's habit choice behavior and the trigger and termination mechanism of habit choice are proposed. Through the simulation experiments, it is found that in the uncertain travel environment, the habit choice behavior of most travelers is a beneficial factor for the system to move towards equilibrium. Finally, on the basis of Swarm platform, the micro simulation system based on multi-agent is designed and developed, and the micro simulation experiment based on the above theoretical model is realized.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：U491.17

【相似文献】