基于浮动车GPS数据的公交系统“盲点”识别方法研究

Blind spots in public transport system are widely appeared in the modern cities, as fast growth of urbanization being out of step with public transport promotion. The uncovered areas of public stations are attribute to the lack of transportation infrastructure, while the inconvenient bus lines are inevitable consequence of flawed transportation planning. These two types of blind spots in public transport system seriously influence the daily commute and other trips. However, way to identify the extents of blind spots are inefficient because the residential trips are changing every day. This project intend to give a definition of blind spot in public transport system from point process theory and present a new method to identify two types of blind spots in public transport system. The classic density-based algorithm will be improved with new definitions of core point and cluster to solve the clustering problem under constraint of incomplete road network. Performance of this method will be assessed by comparing simulated data and clustering result. In other hand, we take Beijing as a case study and analyze spatial distributions and mechanisms of the blind spots, so as to provide a basis for optimization of urban public transport.

随着城市化进程的快速推进，居民出行与公交系统的供需矛盾日益突出，公交系统覆盖外的“盲点”问题越来越普遍，严重影响了居民出行与城市交通。然而，“盲点”的识别问题却一直是一大难题。传统以经验为主的识别方法难以描述“盲点”多时相，不定中心，模糊边界的特征。为此，本项目拟借助点过程理论对公交系统“盲点”明确定义，通过对传统点过程聚类算法中的丛集判定以及模式提取方法改进，提出一个基于出租车GPS轨迹数据识别城市公交系统“盲点”的方法，对公交系统的站点“盲点”、线路“盲点”的中心位置以及空间范围进行识别。研究以北京市为案例，分析了北京市公交系统“盲点”的分布规律与成因机制，从而为城市公共交通优化决策提供依据。

城市化进程的快速推进迫使城市公交系统不断的调整和升级，但依然难以完全满足城市规模扩张与居民出行的多样化需求。这种矛盾体现为因出行需求的不均衡以及其与交通资源的不匹配而产生的出行的热点、冷点、盲点、异常等现象。然而，传统基于静态点分布的分析方法主要研究单一的“来”或“去”特征，割裂了出行流的整体，不足以反映居民出行模式的内在机制传统的，难以去刻画居民的动态出行需求。对此，本课题从点过程基本理论出发，研究了点分布模式的分析方法，然后将点的聚类方法扩展至OD出行流上。以出行OD为核心，构建了基于移动轨迹数据的复杂点过程聚类方法以及流丛集模式识别方法，并将其应用在城市通勤模式识别的应用问题中。结果表明，通过出行流的聚类分析方法，可以有效地发现各类不规则形状的出行流簇以及城市交通中的不同类型的出行模式。该方法以北京市为案例，可以发现城市交通中热点、冷点、盲点、中心点等多种丛聚模式，为城市公共交通规划决策提供有力支持。