[关键词]
[摘要]
为满足城市轨道交通车站精细化客运组织需求,有必要根据车站进出站客流特性进行分类管理。对此,结合AFC采集的进出站客流数据,从车站进出站客流总量及时序特性双角度入手,提出一种基于K-means算法的双层规划聚类方法对全线所有车站进行聚类并划分车站类型。首先以车站进出站客运总量为特征指标进行上层聚类,得出不同客运规模的车站大类;再考虑车站进出站客流的时变特征,根据不同时段内的客流变化特点构建特征向量进行下层聚类,识别车站客流的时序分布特性。通过分析其分类结果与实际高度吻合,不同类别车站在客运规模和时变特性上差异明显。可以看出,所提出的双层k-means聚类分析算法能够很好把握客运规模和客流时变特征,对车站进行精细划分,可为车站的客运组织提供依据。
[Key word]
[Abstract]
In order to meet the demand for meticulous passenger transportation organization in urban rail transit stations, it is necessary to classify and manage stations based on the characteristics of entrancing and exiting passenger flows. Based on the entrancing and exiting passenger flows data collected by AFC, this paper proposes a two-layer planning clustering method based on the K-means algorithm to cluster and classify all stations on the entire line from the two aspects of the total amount of passenger flow and the temporal characteristics of entrancing and exiting passenger flows. Firstly, the upper-layer clustering is carried out based on the total amount of passenger flows of entrancing and exiting as the characteristic indicator, and different types of stations with different passenger transport scales are obtained. Then, considering the temporal characteristics of entrancing and exiting passenger flows, the feature vector is constructed according to the characteristics of passenger flow changes in different periods for lower-layer clustering to identify the temporal distribution characteristics of passenger flow in the station. The classification results are highly consistent with the actual situation, and there are significant differences in passenger transport scale and temporal characteristics among stations of different categories. It can be seen that the proposed two-layer K-means clustering analysis algorithm can well grasp the passenger transport scale and temporal characteristics of passenger flow providing a basis for the passenger transportation organization of stations.
[中图分类号]
[基金项目]
南昌轨道交通集团有限公司科研项目(2021HGKYC005)、江西省教育厅科学技术研究项目(GJJ210654)资助