Changes In Travel Patterns Of Rail Transit Passengers Based On Long-term Smart Card Data

During the long-term development of cities,the travel structure of residents is constantly changing.Due to the booming development of urban rail transit in recent years,its share in public transport continues to increase.More and more people choose to travel by urban rail transit.Although the travel patterns of passengers show a certain stability in the short term,in the long term,the travel patterns of passengers will change due to the influence of various factors.Changes in travel patterns imply changes in passenger demand.In order to improve the efficiency of rail transit systems and reduce congestion,it is necessary to dig deeper into the travel flow characteristics and distribution features of travellers over long time spans.At the same time,automatic fare collection systems are commonly used in urban rail transport,and smart traffic cards have been widely used,which provides conditions for the analysis and mining of passenger travel patterns.Therefore,based on the long-term smart card data of Beijing’s rail transport,this paper digs deeper into the characteristics of passenger travel patterns,identifies changes in individual travel patterns and analyses their changing forms.Specifically,the paper accomplishes the following.Firstly,based on the smart card data of Beijing subway for a total of 13 weeks over four years,long-term travel passengers were selected as the research object,and the travel patterns of passengers were counted in three dimensions,namely frequency,time and space.The K-Means algorithm was used to cluster the passengers and analyse their travel patterns in terms of weekly travel frequency,daily first travel time and daily travel distance.Secondly,using the principle of change point detection,a Bayesian online change point detection model is constructed.And the exact and lagged exact online Bayesian change point detection algorithms in the model were used to detect the travel patterns of individual passengers in three dimensions: frequency,time and space,respectively,to identify changes in passenger travel patterns.The accuracy of the two algorithms for change point detection is compared,and the lagged exact online Bayesian change point detection algorithm,which is more accurate in identifying changes in travel patterns,is used to detect the travel behaviour of long-term travelling passengers in different dimensions,and to analyse the coupling relationship between changes in travel patterns in different dimensions.The change characteristics of different categories of passengers’ travel patterns are analysed for single-dimensional changes and multi-dimensional coupled changes.Finally,the impact of the Beijing subway staggered fare concession policy on the travel patterns of long-term travelers in the time dimension is analysed,using a differences-in-differences method to quantify the extent to which the implementation of two policies with different fare concessions affects passengers’ rescheduling of their travel.This paper constructs a detection model for identifying changes in passenger travel patterns based on long-term smart card data,which can not only accurately detect changes and changing characteristics of individual passengers’ travel patterns in different dimensions,but also quickly identify groups of passengers whose travel patterns change in different dimensions.Therefore,this method can accurately grasp passenger travel and its changing patterns,and then analyse the impact of policy measures on rail transit passenger travel,providing reference for future urban rail transit policy formulation and effect evaluation,passenger flow management and regulation.