Urban Traffic Demand Analysis And Travel Mode Detection Based On Mobile Phone Data

Travel information is essential in conducting traffic planning and traffic management under the complex transportation environment.The traditional methods usually use sample survey results with a sampling rate of within 5%.After that,fixed-point acquisition methods such as coils,RFID,and video are utilized.The former method has long investigation cycle and is highly costed.The result usually depends on the accuracy of the surveyee's memory.The latter method requires the installation of equipment,which has high cost and the data quality depends on the layout of the equipment.Overall,it is difficult to reflect the situation of the entire network.Mobile data has shown tremendous advantages recently.4G technology has greatly improved the quality of mobile phone data.The full-time coverage of mobile phone data is obvious,and it is easy to get full-time user information for the entire region.Besides,mobile phone data comes directly from mobile communications,no additional equipment is required.Based on the above considerations,this paper makes full use of the full-time characteristics of mobile phone data to extract the travel demand information and travel mode of Suzhou.First of all,this paper investigates the current status of urban traffic and the use of mobile terminals.After studying existing traffic data collection techniques,OD estimation,and travel mode extraction methods.This thesis summarizes the deficiencies of existing researches and proposes the direction of the study.Secondly,after reviewing the development of mobile communication,the characteristics of mobile communication technology and positioning technology were briefly introduced.Subsequently,the characteristics of data such as 4G signaling data,triangulation data are described in detail.In the last part of the third chapter,the necessary pre-processing methods for this study are highlighted,which include a base station and traffic cell mapping method and spatial index construction method.When dividing the service range of the base station,this paper uses the Tyson polygon method.Considering the computational efficiency of Hive platform data manipulation,a hybrid spatial indexing method of grid and R-tree is used to divide the study area into equidistant grids and establish multi-level indexes,which greatly reduces the time complexity of subsequent algorithms.Afterward,this thesis shows the example of extracting the traffic production and attraction volume during the early peak period in two traffic zones of Suzhou City.Compared with traditional traffic surveys and manual counting methods,this method greatly reduces manpower and material resources and can obtain more accurate and comprehensive information.Then,based on the 4G mobile phone signaling data,a threshold method is used to estimate the OD information.The trial-and-error method is used to compare the average number of trips obtained by different stay duration thresholds and handover duration threshold,and then compare the results with the trip surveys conducted in Suzhou City in the same year.This method was then applied to the analysis of the employment and residence in Suzhou.The user's place of residence and workplace were studied and the commuting trips between administrative regions were counted.This result can guide regional traffic planning and can play a guiding role in improving road congestion and other issues.After that,the travel mode extraction method was studied based on triangulation data.Firstly,the data cleaning processing is conducted,and the base station of the data source determines whether the travelers traveled by subway.Then,the non-linear threshold and the slow threshold were set to distinguish slow traffic,public transportation,and private car.Different weights are evaluated based on the importance of the mode to extract the dormant mode of travelers.Finally,this thesis summarizes the research results and innovations and puts forward some details that can be improved.This research proposes urban travel demand estimation and travel mode recognition methods,which can accurately describe the current status of urban traffic and accurately portray regional traffic links.