Research On Method Of Temporal And Spatial Distribution Of Urban Travel Demand Based On Mobile Phone Data

The traditional travel survey methods are difficult to obtain the real-time travel information due to high cost,long period and the small sizes of survey sample data,etc.With the rapid development of location sensing technology and the penetration of mobile communication networks in major parts of urban areas,abundant mobile phone data could be provided data support for the study of the characteristics and demands of residents' travels.Data mining,machine learning and other related algorithms can be used to analyze mobile phone data.And residents' travel information is obtained.A complete set of quantitative methods for analyzing traffic demand is established.Based on this phenomenon,this paper puts forward a study on the distribution of travel demands.On the one hand,the method on extracting traffic OD(Origin and Destination)is applied.On the other hand,travel demand is analyzed.The procedures are as follows:Firstly,according to generation mechanism of original mobile phone data,signaling trigger frequency and temporal and spatial characteristics of data is analyzed.The methods of data preprocessing are proposed based on the characteristics of and requirements on data.Invalid data such as redundancy,static data,ping-pong data and virtual position are removed.High quality data are obtained for follow-up researchSecondly,based on the tidal variation of mobility in base station and classification of land use,the volumes of users,aggregation,dissipation and new increment are analyzed.The clustering algorithm(K-MEANS)is used to identify urban functional areas.Temporal and spatial characteristics of base stations and traffic semantics are chosen to divide traffic zones by using weighted Mahalanobis distance algorithm.The commuting index is defined to measure the tendency of traffic zones on jobs-housing attribute.Thirdly,the characteristics of trajectory are analyzed.According to the three constraints of time,space and frequency threshold,the judgment rules of traffic OD are determined.And then the temporal and spatial OD model is established.The trajectory points are divided into the static and the moving,simultaneously the OD data are obtained based on the static points.The F1-MEASURE method is used to verify the accuracy of clustering algorithm.Finally,based on the extracted OD data,a calculation method is used for travel indexes including travel distance and travel time,etc.Compared with traditional surveys,the results of travel characteristics have a higher degree of reliability.The evaluating state of road network is analyzed by OD distribution map,population density map and urban tidal movement map.which could provide reliable support for the running state of road network.