Research On Big Data Visualization And Prediction Of Urban Rail Transit Network Passenger Flow

The vigorous development of urban rail transit has made it necessary to process large scale and various types of data in passenger flow research.With the continuous improvement of information technology,passenger flow data has been collected and stored effectively while increasing rapidly,providing basic data resources for the analysis of passenger flow characteristics and the research of passenger flow prediction of urban rail transit.This thesis uses big data technology for mining and analysis to obtain the distribution information of passenger flow characteristics by station,line,time and region hidden behind the massive subway passenger flow data,and integrates the trend law of passenger flow with the geographical location information of line stations to achieve visual and dynamic presentation,and intuitively grasp the spatial and temporal distribution law of urban rail transit passenger flow from the disorganized data.Understand the status of key indicators of road network operation.The passenger flow prediction model is constructed and applied to the actual traffic system to help the passenger flow guidance and control.It is of great practical significance and application value to help passengers timely understand travel road conditions,avoid congestion periods and congestion stations,rationally plan residents’ travel behaviors,and help the development of the industry.The main research achievements and work of this thesis are as follows:(1)In view of feature mining of network passenger flow big data,subway card swipe data,geographic information data of subway network and point of interest data of subdistricts around the city center were acquired for data fusion and preprocessing.The travel demand and travel behavior of passengers are analyzed,and the influence mechanism of subway passenger flow is studied,which lays a theoretical foundation for the analysis and visualization of subway passenger flow characteristics and the prediction content.(2)In view of the visualization of big data of network passenger flow,the basic data model is established to solve the problem of data form mismatch,and the regional dynamic OD migration map of integrated GIS is output,such as the distribution diagram of line cobweb distribution of upstream and downstream sections of passenger flow,the comparison diagram of passenger flow distribution in and out of the station,and the spatial and temporal distribution diagram of multi-view fusion,etc.The visual dynamic display of passenger flow spatial-temporal fusion distribution characteristics and dynamic synchronization of passenger flow data spatial-temporal information can be realized to present the state of variables such as travel time and spatial location that passengers care about in an intuitive form.(3)Study the big data prediction of network passenger flow,and use the geographical weighted regression(GWR)and multi-scale geographical weighted regression(MGWR)models to explore the interrelationship and dependence degree between different types of POI and each station generating passenger flow.The direct demand prediction model of subway passenger flow based on time series(ARIMA)is constructed,and the actual passenger flow data is used as the training data to build the prediction model,and the short-term passenger flow prediction is realized.(4)Design and develop the subway network passenger flow big data feature analysis system,receive traffic network and equipment data in real time through various systems and terminals,realize multi-view fusion display of passenger flow feature distribution through the front end,analyze access data in real time,realize interactive visualization of analysis results,and realize system functions by modules.