Metro Passeger Flow Analysis Based On Multisource Data

Rail transit system is an important part of urban public transport system.By reinforcing the planning and operation practice of the rail transit system,the willingness of citizens choosing public transit to travel will increase,thereby relieving the pressure on other parts of the entire transport system and relieving traffic congestion.Accurate acquisition of passenger flow of the rail transit system is of great reference significance to the optimization of current lines and the planning and adjustment of new lines.Traditional researches of rail transit passenger flow based on smart card data can only obtain the spatio-temporal distribution characteristics of passenger flow within the system,but cannot accurately obtain the passengers’ complete travel trajectory information.With the continuous popularity of mobile phones and development of communication technology,the usage of mobile phone data in the analysis of urban residents’ travel are increasing rapidly.As base stations are widely spread in the city,and special base stations are installed underground at stations and along the routes,we can extract the whole track of passengers by analyzing mobile phone signaling data.Based on the full trip chain information,the distribution characteristics of the origin and destination of these passengers outside the rail transit system can be derived,thereby providing reliable data support for the evaluation of the service range of existing lines,the adjustment of existing lines and planning of new lines.Based on the mobile phone signaling data,this study extracts the travel chain related to rail transit passengers,constructs the evaluation framework of rail transit system passenger flow on both spatial and temporal scope,and builds a passenger flow prediction model of rail transit system stations.At beginning,this paper reviews current passenger flow analysis methods of rail transit based on credit card data,and investigates the existing studies on the usage of mobile signaling data in the analysis of rail transit passenger flow.By summarizing the existing subway passenger flow prediction methods,it demonstrates the possibilities of using mobile phone data in station passenger flow prediction tasks.Based on the results of literature review,the direction of this research is determined,that is,the construction of the analysis and prediction system of rail transit passenger flow based on fusing of different data sources such as mobile phone signaling data and smart card data.After a brief introduction to the history of mobile communication technology,this paper introduces all kinds of data sources used in the research,including mobile phone signaling data,smart card data,GIS data,weather data,etc.Because the original mobile signaling data contains a lot of noise,chapter II introduces the noise eliminating method of ping-pong data and drift data contained in the mobile signaling data in detail.After removing the noise in raw data,the path of each traveler can be represented by sequence of base stations more accurately.Based on the preprocessed data,by using spatiotemporal characteristics of the passengers’ travel chain,this paper introduces an identification method of traveler’s on-board and off-board station.At the same time,consider that mobile phone signaling data can track the whole process,combined with the topological characteristics of rail transit line network,this paper put forward a method to identify the passengers’ transfer station.The results could be used to analyze the characteristics of different transfer lines.In the last part of the chapter III,based on the smart card data,this paper also verifies the results of traveler track identification.The results show that the rail transit travel quantity obtained based on the mobile phone data has a high accuracy.The methods mentioned above could only derive the passengers’ trajectory inside the rail transit system,considering the feature that mobile signaling data can record the passenger’s trajectory outside the rail transit system,this paper introduces an identification method of the passenger’s origin and destination,and constructs an evaluation framework for analyzing the results.The method proposed in chapter IV can be used for deriving the spatial and temporal distribution of the origin and destination of each line and each station,and then extract the service range and other characteristics of each station.Passengers’ complete trajectory could only be used for macroscope such as network planning.To further provide more solutions in operational scopes,by integrating the inbound passenger flow characteristics obtained from mobile phone data with the outbound passenger flow characteristics obtained from smart card data,along with other input features,this paper builds a deep learning framework for short-time passenger flow prediction of the station,achieving high prediction accuracy.Finally,the main results of this study are summarized,along with research directions in the future.With the proposed rail transit passenger flow analysis method,the internal and external trajectory of rail transit passengers,and the overall passenger flow characteristics of rail transit system could be precisely depicted.And the results could provide strong support for the authorities on both macro and micro tasks.