VMS信息作用下的驾驶员路径选择行为研究

发布时间：2018-06-04 02:30

  本文选题：诱导信息 + 路径选择　； 参考：《天津大学》2015年博士论文

【摘要】：交通流的产生离不开出行者的选择和决策,其本质就是出行者出行行为集计的结果,而路径诱导系统的效果归根到底也取决于出行者在诱导信息作用下的路径选择行为,所以说诱导信息作用下路径选择行为的研究具有重要的理论意义和现实价值。本文首先通过调查统计和模拟实验研究驾驶员路径选择行为的特点和决策机制,并以此研究结论为基础在认知心理学框架下提出真实合理的行为假设,构建符合实际的行为决策模型来模拟和分析出行者的路径选择行为,最后采用多智能体仿真技术研究交通信息的发布模式,为交通诱导系统的构建和诱导信息的发布提供理论依据和技术支撑。本文的工作内容和创新点汇总如下:1.设计并通过SP场景实验获取VMS信息作用下驾驶员路径选择的行为数据,通过经验路径、最优路径和选择路径之间的关系定义了驾驶员对诱导信息的服从行为,并基于服从率构建有序回归模型,总结了影响驾驶员服从信息的不同因素,结果表明年龄、收入、性格、职业、对信息的信任程度、出行经验和路径选择风格直接影响诱导信息对驾驶员的作用。2.基于前景理论对驾驶员路径选择行为中的风险态度进行分析,结合问卷和调研的实证数据对风险态度参数进行估计并验证。同时考虑了驾驶员间的异质性,得到不同类别驾驶员的风险态度特征,结果表明随着年龄的增加,驾驶员间的异质性更加明显,而且在面对收益时的风险规避程度随着收入的增高而下降。3.开展模拟实验研究诱导信息作用下出行者的逐日路径选择行为,探究驾驶员从内部经验和外部信息进行学习的过程,研究发现随着出行经验的增加,出行者对不同路径会形成偏见,路径选择决策中习惯作用比较明显,习惯的作用随着决策环境复杂性的增加而增大。4.借鉴认知心理学和SOAR认知框架,基于对调查、模拟等数据研究获得的结论和启示对路径选择行为的认知过程进行描述,同时反映驾驶员在路径选择行为中的感知、工作记忆、长期记忆、组块和强化学习等功能。5.对比分析多智能体仿真模拟结果和调查统计获取数据,结果表明基于SOAR认知框架的路径选择模型能够合理有效地模拟现实决策过程,并基于此认知模型仿真得到预测信息发布模式在诱导效果和路网均衡上优于实时信息模式。
[Abstract]:The generation of traffic flow is inseparable from the traveler's choice and decision, and its essence is the result of the traveler's travel behavior set, and the effect of the path guidance system in the final analysis also depends on the traveler's path choice behavior under the action of the induced information. Therefore, the study of path selection under the action of induced information has important theoretical and practical value. In this paper, the characteristics and decision-making mechanism of driver's path choice behavior are studied through investigation, statistics and simulation experiments, and based on the conclusions, a real and reasonable behavior hypothesis is put forward under the framework of cognitive psychology. A practical behavior decision model is constructed to simulate and analyze the path selection behavior of travelers. Finally, the multi-agent simulation technology is used to study the traffic information release model. It provides theoretical basis and technical support for the construction of traffic guidance system and the release of guidance information. The contents and innovations of this article are summarized below: 1. The behavior data of driver's path selection under the action of VMS information are designed and acquired by SP scene experiment. The behavior of driver's obedience to induced information is defined by the relationship among empirical path, optimal path and selective path. An orderly regression model is constructed based on the rate of serviceability, and the different factors influencing the information of drivers' clothing compliance are summarized. The results show that age, income, character, occupation, degree of trust in information, etc. Travel experience and path selection style directly affect the effect of induced information on drivers. 2. Based on the foreground theory, this paper analyzes the risk attitude in the driver's path choice behavior, and estimates and verifies the risk attitude parameters based on the questionnaire and empirical data. At the same time, considering the heterogeneity of drivers, the risk attitude characteristics of different types of drivers are obtained. The results show that the heterogeneity between drivers is more obvious with the increase of age. And the degree of risk aversion in the face of income increases as income increases. 3. 3%. The simulation experiments were carried out to study the daily path selection behavior of travelers under the action of inducing information, and to explore the learning process of drivers from internal experience and external information. The study found that with the increase of travel experience, Travelers have a bias against different paths, and habit plays an obvious role in path selection, and the role of habit increases with the increase of decision environment complexity. Using the cognitive psychology and SOAR cognitive framework for reference, the cognitive process of path selection behavior is described based on the conclusions and revelations obtained from the investigation, simulation and other data studies, and the perception and working memory of drivers in the path choice behavior are reflected at the same time. Long-term memory, block building and reinforcement learning. 5. The results show that the path selection model based on SOAR cognitive framework can simulate the real decision-making process reasonably and effectively. Based on this cognitive model, the predictive information release model is better than the real-time information model in the inductive effect and road network equilibrium.
【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
，

本文编号：1975438