Study On Characteristics And Influencing Factors Of Urban Public Transport Passenger Flow In Different Scales

With the rapid development of urbanization,great pressure of the sustainable development of urban environment and society is approached.Urban public transportation is an important means to realize sustainable development of transportation and promote the social equity,which is the research hotspots.The characteristics of urban public transport passenger flow are the interactive results of public transport stations and their surrounding urban space.Accurate identification of the passenger flow characteristics of urban public transportation and analysis the influence factors are the premise of rational planning of transport facilities,effective improvement of service efficiency and strengthening of comprehensive development orientation.The improvement of smart card system and the enrichment of urban big data provide a more comprehensive and accurate way to study the relationship between urban public transport passenger flow characteristics and urban space quantitatively.However,the current research about the definition of passenger-flowtaking-place space is vague.At micro level,station area is defined as a single radius buffer zone,and at meso level traffic analysis zone(TAZ)is often considered to be the basic unit of traffic investigation.Obviously,it is difficult to truly and comprehensively reflect the characteristics of urban public transport.The common methods of passenger flow extraction have some limitations on mining highdimensional and complex sequence data.A feasible and robust data mining method is needed to obtain passenger flow features effectively from high-dimensional sequence data.In addition,most of the existing studies focus on the impact factors of built environment on passenger flow characteristics in a single scale,ignoring the differences of the impact factors of the overall urban space in different scales.In view of the existing shortcomings,based on the smart card data,geographical conditions data,land use data,POI data,population data and other multi-source city big data,this thesis first constructs the basic unit for spatial analysis at different scales,and proposes improved affinity propagation(AP)algorithm to recognize the characteristics of passenger flow at different scales;then analyzes the influence factors of the total passenger flow and the elderly passenger flow in station area from the perspective of walking accessibility at the micro level;finally,the thesis analyzes the influencing factors of the total passenger flow and the elderly passenger flow in TAZ from the perspective of travel similarity at the meso level.The main research contents and conclusions are as follows:(1)The station area unit based on walking accessibility and TAZ unit based on travel similarity represent passenger-flow-taking-place space at different level.On the one hand,this study uses the Amap web service to obtain the walking time cost of the station and its surrounding grid,then obtain the station area based on walking accessibility.The results reflect the real range of passenger flow activities more accurately and effectively,and provide an basis for exploring the relationship between passenger flow characteristics of public transport stations and the urban space and population around the stations.On the other hand,this study proposes a TAZ division method based on travel similarity,which consists of the traditional division principle of "partition" and the data-driven method of "aggregation".The method reduces the complexity of the data-driven method,improves the efficiency and readability of the division results.LB?Keogh distance is the way to measure similarity of the grid unit,which considers the characteristics of urban public transport effectively.The results show that the scheme is more reasonable than the traditional partition scheme and the classical data-driven partition scheme,and the difference of area and trip density between TAZ units is smaller.This method,which takes into account the comprehensibility of cognitive level and rationality of data level,provides a feasible method and research example for "customized" TAZ.It provides a basis for identifying passenger flow characteristics and analyzing its influencing factors in TAZ scale.(2)This thesis proposes an AP clustering algorithm based on Dynamic time warping(DTW)distance to mine high-dimensional and complex passenger flow sequence data.Without prior knowledge,it can effectively recognize the characteristics of the total passenger flow and the elderly passenger flow the in four scenarios directly from the smart card data.AP algorithm based on LB?KEOGH distance can effectively identify the total travel passenger flow and the elderly travel passenger flow in TAZ scale.The results show that there are significant differences in the quantity and time of public transport stations,and the time series characteristics mainly include "double peak type","single peak type","step down type" and "stable type".The results show that the passenger flow of rail transit is larger than that of bus,and there is a phenomenon of "axis shifting" in time characteristics;the number of elderly people who travelled by rail transit on weekends is reduced,and whether the bus travel is reduced depends on whether the area is close to the business district.Under the TAZ scale,the total passenger flow shows a "double peak type" time series characteristic,while the elderly passenger flow shows a "step down" time series characteristic,and there are also significant differences in the quantity.TAZ unit with large passenger flow of public transport accounts for only 10%,and they are distributed along the rail transit in the central urban area.This method effectively solves the limitation of mining of unknown facts in existing methods,further researches the public transport characteristics for the elderly,and provides a reasonable reference for adjustment of facilities and policies for the elderly and the improvement of "people-oriented" service quality in the future.(3)Explore the scale effect of influencing factors of passenger flow characteristics.On the basis of large-scale geographical data,land use and POI data are fused,and urban functional landscape system is constructed by re-classification as the main influencing factors.A multiple logistic regression model is established to explore influencing factors of passenger flow characteristics at different scales.The results show that the impact of urban functional landscape on urban public transport passenger flow has significant scale effect.In the socio-economic functional landscape,transportation,commerce,industry,leisure,residence,education and medical treatment have significant effects on different passenger flow characteristics at different station scales,while education and medical treatment have significant effects at TAZ scale.Potential urban functional landscape proportion was introduced and find that it has a significant negative impact on the 5-minute and 10-minute scale of the station area,and has a greater impact on the passenger flow of rail transit stations than on the bus stations,especially on the rail transit for the elderly;it has no significant impact on the TAZ scale.At the landscape level,Patch Richness(PR)and Interspersion Juxtaposition Index(IJI)have no significant impact on rail transit passenger flow,but have a significant positive impact on bus passenger flow.In addition,the influence of population density on the passenger flow of bus stops is larger than that of rail transit stops on the scale of station area,while the influence of population density on the scale of TAZ is significant.The differences of influencing factors at different scales provide effective reference for comprehensive utilization planning of public transport landplanning at micro and meso levels.The above results show that the basic analysis unit division of different scales considering the real urban road network and public transport characteristics can more reasonably reflect the spatial location of passenger-flow-taking-place space,and provide the basis for the analysis of passenger flow spatio-temporal characteristics and its influencing factors;the AP clustering algorithm based on DTW distance can more objectively and effectively identify high-dimensional and complex data without subjective influence of prior knowledge;the urban functional landscape system can describe urban functional space more precisely and comprehensively,and has significant scale effect on urban public transport passenger flow.This study provides a useful reference for the orderly management of urban public transport,the coordinated development of urban functional space and the comprehensive utilization of urban public transport land.