时空约束下的微观出行网络模型及演化机制研究

The intercross of multi-subjects and the study based on micro-behavior model have become the research trend of traffic demand modeling in recent years. However, there are deficiencies in that the current research lack deep analysis on the characteristics of micro-behaviors as well as their relationships under spatio-temporal constraints. Therefore, the theories of complex network and time-geographic are proposed to discuss the microscopic travel network modeling under spatio-temporal constraints, and explore the evolution of travel communities as well as the dynamics spreading processes of space-time bundles in this study. Firstly, a micro-travel oriented network, which has double layers, is established to explore the activity-travel patterns of people in a three-dimension environment under spatio-temporal constraints. Then, by introducing the dynamic and the static characteristics of community evolution, an algorithm is designed to detect the structures of travel communities and reveal their evolution rules in large-scale travel network. Furthermore, a quantitative analysis of the dynamics spreading process of space-time bundle is conducted considering the characteristics of traveler's adaptive avoidance and travel network's open growth, while the research on lagged response mechanism of traffic management policy is also carried out under the theory of bounded rationality. This study is expected to result in a set of methods and tools that can be used to enrich and extend the existing theories and methods in microscopic travel behavior research field, systematically analyze urban traffic system problems, and formulate effective traffic management strategies.

近年来，多学科交叉和以微观行为分析为基础的模型研究成为了交通需求建模的发展趋势。然而现有研究对时空约束下的微观行为特征，以及出行活动之间的联系缺乏深入分析。因此，本项目拟结合复杂网络理论和时间地理学框架，研究时空约束下的微观出行网络建模、社团结构演化，以及时空束传播等问题。首先探索三维时空约束下居民活动和出行的表达方法，建立面向微观出行的双层网络模型；然后从社团结构演化的时序动态性和时刻静态性入手，设计适用于大规模出行网络的社团探测算法，揭示出行社团的演化规律；进而考虑出行者的自适应躲避和出行网络的开放式增长等特点，定量研究时空束传播的动力学过程；并以有限理性决策理论为指导，开展出行社团对交通管理政策的滞后响应机理研究。研究成果将丰富和拓展微观出行行为研究的已有理论和方法，为诊断城市交通问题、制定和实施高效的交通管理策略提供科学有效的分析方法和工具。

本项目在北京、上海、和哈尔滨分别选取高速发展的城市核心区域作为研究区域，实现了全方位多源交通数据的采集。面向采集的多源异构交通信息，提出了基于模糊C均值的数据质量控制方法，提出了一种基于相空间重构的交通流复杂网络建模方法，利用复杂网络理论分析了不同采样间隔下和不同交通状态下的交通流特征，并提出了基于双指数平滑和支持向量机的组合交通流预测模型，有效的实现了面向多源数据的交通流知识提取和模式识别。基于空间点模式分析技术和动态核密度估计理论，构建了居民出行讫点预测模型，分析了城市内居民出行讫点的时空分布规律，进行了出行热点区域探测，以上研究成果对于公共交通的有效协调和组织有着重要的指导意义。面向我国交通调度中常见的公交串车问题，建立了公交车运行状态的预测模型，进而提出了以车均延误最小和乘客感知偏差最小为优化目标，结合信号调度和滞站策略的控制方式建立了多交叉口协调控制模型。上述研究成果丰富和拓展了微观出行行为研究的已有理论和方法，对于完善和提升交通需求分析方法具有重要的理论意义。