基于地铁客流与建成环境映射关系的新线车站分类方法

A classification method for new line stations based on the mapping relationship between subway passenger flow and built environment

科学合理的车站分类对于客流特征研究预测、车站设施布局优化以及周边土地开发建设具有重要意义。针对新线开通前新建车站客流数据未知而导致车站分类不准的问题,提出一种基于地铁客流与建成环境映射关系的新线车站分类方法。首先,从客流角度选取组合聚类指标对既有车站进行聚类,形成各类车站的客流特征区间。其次,从建成环境角度计算既有车站的土地混合熵、不同POI(Point of interest)数量、度和介数等特征,利用随机森林(Random forest,RF)重要度指标筛选和对数变换拟合确定建成环境特征与地铁客流特征的映射关系表达式。最后,输入新线车站的建成环境特征,依据上述映射关系计算新线车站的客流特征值,结合既有车站的客流特征区间进行新线车站客流特征归类,实现客流数据缺失条件下的新线车站分类。为验证方法有效性,采用北京地铁数据进行案例验证。研究结果表明:基于地铁客流与建成环境映射关系的分类方法能够实现新线车站开通前的车站类型提前划分,具有较好的分类结果;进一步分析发现工作日车站与周末车站分类存在的差异与通勤行为特性、土地开发程度有关;研究方法在不同线路数据集上都表现出较好的分类效果,具有较强的适用性。研究结果可为缺失数据条件下的车站分类和新线开通期地铁运营组织提供新的思路和方法。

Reasonable station classification is of great significance for the research and prediction of passenger flow characteristics,the optimization of station facilities layout and the development and construction of surrounding land.To solve the problem of inaccurate station classification due to unknown passenger flow data before the opening of new line,a new classification method for new line stations was proposed according to the mapping relationship between the built environment and subway passenger flow.Firstly,the existing stations were clustered by mixed clustering indexes which were selected from the passenger flow perspective,and the passenger flow characteristic interval of each kind of station was calculated.Secondly,from the built environment perspective,features such as land mixed entropy,the number of different points of interest,degree,and existing stations were calculated.Random forest importance index screening and logarithmic transformation fitting were used to determine the mapping relationship between built environment characteristics and subway passenger flow characteristics.Finally,according to the above mapping relationship,the passenger flow characteristic value was calculated by the inputting the built environment characteristics of the new line station,and compared with the passenger flow characteristic interval of the existing station,the classification of the new line station under the condition of missing passenger flow data was realized.To verify the effectiveness of the method,Beijing subway data was used for case verification.The research results show that the classification method based on the mapping relationship between subway passenger flow and built environment can achieve the classification of station types in advance and better classification results before the opening of new line stations.Further analysis shows that the differences in the classification of weekday stations and weekend stations are related to commuting behavior characteristics and land development levels.The present method has strong applicability and performs well on different line data sets.The research results can provide new ideas and methods for station classification under the condition of missing data and subway operation organization during the opening of new lines.