基于EM算法与Gauss混合模型的地铁站点类型分析

结合成都地铁 2019 年 5 月工作日的 AFC 数据，利用 EM 算法(expectation-maximum algorithm)与 Gauss 混合模型，分析成都市 156 个轨道站点客流曲线的特征差异，并结合平方和误差将其聚类为居住导向型、就业导 向型、职住错位型、错位偏居住型、错位偏就业型、交通枢纽型、综合型 7 种不同类型的地铁站，最后分析不同 类型地铁站的区域分布及土地性质。研究表明，不同类型站点分布具有区域性，站点类型随着到城市中心距离的 增加而减少，中心区站点类型更多样，可体现城市功能区域时空差异的表现形式，提供城市空间进行研究的新视 角，有助于了解城市功能的空间分布，为未来城市及交通规划提供依据。

Type Analysis of Metro Stations Based on EM Algorithm and Gauss Hybrid Mode

In this study, we analyzed the differences in passenger flow curves using the expectation maximization algorithm and Gauss mixed model. The automatic fare collection (AFC) data of working days in May 2019 of 156 metro stations in Chengdu were used. The stations were divided into living-oriented, career-oriented, living-employment dislocation, partial living dislocation, partial employment dislocation, transportation junction, and synthesis types. Finally, the correlation between the land types and station distributions of different metro station types was analyzed. The results show that the distribution of different stations has regional attributes. The number of metro types decreased when moving further away from the city center. The types of stations in the central area are complex, which reflects the manifestation of spatiotemporal differences in urban functional regions. This study provides a new perspective for the study of urban spaces, helping to understand the spatial distribution of urban functions as well as providing a foundation for future urban and transportation planning.