首页|基于机器学习空间聚类的出租车停靠站点布局规划

基于机器学习空间聚类的出租车停靠站点布局规划

扫码查看
原文链接
  • 国家科技期刊平台
  • NETL
  • NSTL
  • 万方数据
  • 维普
针对出租车随意停靠给城市交通带来的负面影响,为规范出租车营运秩序、改善出租车营运环境和居民乘车条件,提出一种将出租车出行空间信息与机器学习算法相结合的出租车停靠站点布局规划方法.首先利用出租车GPS轨迹数据提取出租车出行起点,然后采用HDBSCAN聚类算法对起点进行空间密度聚类,形成聚类簇后以其中心点作为出租车停靠站点布局的备选点.最后,为验证所提方法的可行性和有效性,选取重庆市中心城区一土地利用类型丰富、人口密度高的典型区域进行案例分析.结果显示,107个备选点主要分布于商业中心区和居住集中区,与出租车出行高需求区域的空间分布基本吻合;布局的出租车停靠站点在300 m范围内的覆盖率达到76.0%,未覆盖区域主要为城市绿地和水体.研究表明,机器学习算法可实现出租车停靠站点的高效布局规划,但在规划和实施阶段,停靠站点的设置还应结合邻近区域的建成环境特点综合考虑.
Taxi Stand Layout Planning Using Machine Learning-Based Spatial Clustering
The arbitrary stopping of taxis has caused a certain negative effect on urban traffic.In order to regulate the order of taxi operation,improve the conditions of taxi operation and residents'riding,a taxi stand layout planning method which combined the spatial information of taxi trips with machine learning algorithms was proposed.Firstly,the GPS trajectory data of taxis was used to extract the origins of taxi trips.Then,the HDBSCAN clustering method was used to perform spatial density clustering on the ori-gins of taxi trips,the clusters were formed and their centers were used as alternative locations for the layout of taxi stands.Finally,to verify the feasibility and efficiency of the proposed method,a typical area with rich land use types and high population density in the central urban area of Chongqing was selected as an example for case analysis.The results showed that the 107 alternative locations were mainly located in commercial centers and residential areas,which was basically consistent with the spatial distribution of areas with high taxi demand.The 300-meter coverage rate of taxi stands in the layout reached 76.0%,and the uncovered areas were mainly urban green spaces and water bodies.Re-search has shown that machine learning algorithm can achieve efficient layout planning of taxi stands,but in the planning and implementation stages,the setting of parking space should also be comprehen-sively considered in conjunction with the characteristics of the built environment in adjacent areas.

urban trafficlayout planningspatial clusteringtaxi standtrajectory datamachine learning algorithmHDBSCAN(Hierarchical Density-Based Spatial Clustering of Applications with Noise)

年光跃、黄建云、潘海啸

展开 >

同济大学 建筑与城市规划学院,上海 200092

上海交通大学 设计学院,上海 200240

城市交通 布局规划 空间聚类 出租车停靠站点 轨迹数据 机器学习算法 HDBSCAN

国家自然科学基金区域创新发展联合基金

U20A20330

2024

10.16503/j.cnki.2095-9931.2024.01.002

交通运输研究
交通运输部科学研究院

交通运输研究

CSTPCD
影响因子：0.941
ISSN：1002-4786
年,卷(期)：2024.10(1)
  • 25