Analysis Of Characteristics Of Urban Individuals’ Travel Behavior Based On Multi-source Big Data

In recent years,the study of residents’ travel/activity behavior characteristics has been a hot topic in the fields of traffic engineering and urban geography.On the one hand,the research of spatial-temporal characteristics of travelers can help urban transportation service providers adjust the resource layout.On the other hand,it can help relevant government departments formulate reasonable traffic planning and management measures to alleviate urban traffic congestion.With the rise of big data technology,it has become possible to study the characteristics of travelers on city’s level.The algorithm designed in this paper realizes the identification of individual travel and joint travel behaviors.The mobile phone data and metro card data are combined to analyze the characteristic of residents’ trips.Specifically,firstly,a rulebased method is proposed for identifying individuals’ occupation and residence distribution based on mobile phone data.Furthermore,the algorithm is designed to identify individuals’ trip chain.Then,the temporal and spatial distribution characteristics of travelers can be studied.Based on the study of individuals’ travel behavior,this paper further expands the study of the spatiotemporal characteristics of the behavior of individuals’ traveling together(referred to as joint travel).For the extraction of joint travel behavior,this paper designs a social relationship classifier to classify the relationship between travelers into acquaintances and non-acquaintances.With the social relationship result,a rule-based method is proposed to recognize whether two acquaintances are traveling together.This paper summarizes the individuals’ joint travel/activity behavior into four different patterns,and uses the metro card data to discuss the temporal and spatial distribution characteristics of the four different joint travel/activity patterns(JATPs).Finally,this paper introduces a network equilibrium model that considers the individuals’ joint travel/activity behavior,and gives out a parameter calibration method of the model.The main results of this paper are summarized as follows:(1)Mobile phone data is suitable for studying the distribution of supply and demand of residents under multi-level time and space scales.In this paper,a rule-based algorithm is used to recognize residence and work place of residents in the research time period.A case study conducted in Suzhou City shows that the algorithm proposed in this paper has good feasibility and accuracy.The calculation results can provide data support for microscopic and macroscopic transportation planning.(2)The mobile phone data can be used to calculate the traffic distribution of urban residents.This paper first calculates the residence time of residents at all base stations and then calculates the travel distribution among different regions by statistics of different time and space levels.The integration of mobile phone data and metro card data at the subway station level not only calculates the traffic flow accurately but also finds the original and destination of travelers.The results can be applied to the study of urban road network traffic distribution,excavating congested road sections,and calculating the distribution of residents’ source and destination places.(3)Residents’ joint activity/travel behaviors are recognized into four patterns.Travel similarity and travel regularity are used to divide residents into acquaintances and non-acquaintances.A rule-based algorithm is used to identify different joint activity/travel patterns(JATPs).The results of case study conducted in Suzhou show that the spatial-temporal characteristic of JATPs is different.This difference is affected by travel time,activity time,and land use of metro stations.(4)A two-stage parameter calibration method is proposed for the network equilibrium model considered JATP.In the first stage,an objective function with the smallest sum of square deviations between the network traffic calculated by the model and the network traffic obtained from the data is established,and the simulated annealing algorithm is used to solve the problem.In the second stage,a Kalman filter is incorporated to reduce the influence of the observation error or system error of the input data on the accuracy of the parameter calibration result.A case study in a small metro network in Suzhou City shows that the parameter calibration algorithm has good convergence and can be further extended to other traffic allocation models.