基于时间序列聚类算法的城市轨道交通节假日客流波动分析

Analysis of Passenger Flow Fluctuation of Urban Rail Transit during Holidays Based on Time Series Clustering Algorithm

与平常相比,节假日期间城市轨道交通客流会出现大幅度的波动,对波动规律进行分析和预测,是城市轨道交通运营管理部门制定客流组织方案的先决条件。鉴于传统方法对样本规模依赖程度较大这一局限性,提出一种基于时间序列聚类算法的城市轨道交通节假日客流波动分析方法。借鉴相似法思想,以相似日为参照,在节假日客流波动率序列和节假日客流波动率差分序列的基础上构建了节假日客流波动描述向量,并以描述向量为依据,实现特征提取和聚类。以重庆轨道交通为研究对象,对3天节假日客流数据进行分析,客流量平均相对误差仅为4.56%。为更加准确地分析节假日客流量,进而为城市轨道交通运营管理部门制定节假日运力调配和车站客流疏导等方案提供了有力的决策支持。

Compared with ordinary days,the passenger flow of urban rail transit fluctuates greatly during holidays.Analyzing and forecasting the fluctuation principle is a prerequisite for the urban rail transit operation management department to formulate the passenger flow organization plan.In view of the limitation of the traditional methods which relied heavily on the sample size,a method for analyzing the fluctuation of passenger flow of urban rail transit during holidays based on the time series clustering algorithm was proposed.Referring to the idea of similarity method and taking similar days as a reference,the description vector of passenger flow fluctuation during holidays was constructed on the basis of passenger flow fluctuation series and passenger flow fluctuation difference series during holidays.The feature extraction and clustering were realized based on the description vector.Taking Chongqing rail transit as the research object,the average relative error of passenger flow is only 4.56% by analyzing the passenger flow data of 3-day holidays.Such a method provides a strong decision support for the urban rail transit operation management department to analyze the passenger flow during holidays more accurately,and further to formulate plans for capacity allocation and passenger flow diversion in stations during holidays.