Metro Passenger Flow Forecasting With Addition Of New Line Considering Land Use Characteristics And Accessibility

Recently,urban rail transit has been developing rapidly in China,and the speed of development does not match the level of management and operation,with the construction and use of new lines,the network topology of rail transit network changes,and the distribution of passenger flow on the network is more and more complex.As an essential link in the early stage of the efficient and safe operating organization,passenger flow demand forecasting determines the service level of the rail transit system,so accurately predicting passenger flow demand is a problem that needs to be solved urgently.After analyzing the deficiencies of existing studies in data acquisition and prediction accuracy,as well as considering the impact of easily accessible land use indicators and accessibility indicators determined by location conditions on passenger flow,this study proposes a metro passenger flow forecasting model that can be applied under the condition of new line access.First,in order to distinguish the passenger flow differences caused by station attribute characteristics and location conditions,a refined land use database,based on web crawler technology,was built,and,based on K-means clustering method,cluster analysis on existing stations was performed.In order to determine the impact of the range of passenger flow attraction on passenger flow,the direct and indirect attraction ranges are determined by the difference in the traffic modes of the connecting stations and by the higher principle.Based on the agglomeration effect theory and the generalized cost model,the model is built to calculate the direct and indirect attraction range of the stations.To avoid possible redundancies in areas where passenger flow attraction ranges overlap,the accurate division of overlap area of passenger flow attraction ranges is achieved based on Tyson polygon theory.Second,in order to capture the impact of rail transit induced passenger flow caused by the improved accessibility of the rail transit system after the new metro line is connected to the network,a destination selection model is established based on the physical characteristics of the rail transit topology network nodes,and the accessibility index of rail transit network is defined based on the LOGSUM utility function so as to quantify the improvement of the accessibility of each station of the line network after the new line is connected.Third,in order to capture the unconventional change characteristics of passenger flow under the new line access and achieve accurate passenger flow prediction,the passenger flow prediction models for existing line and new line are proposed,metro passenger flow for existing stations is deconstructed into three parts: natural change passenger flow,induced passenger flow,and transfer passenger flow.Meanwhile,based on the seasonal ARIMA model and the radial basis neural network,the natural change passenger flow is predicted and the applicability of the two methods to the passenger flow prediction under specific conditions is measured.Based on the nonlinear analysis of the historically opened new line passenger flow data,this paper obtains the quantitative weight coefficients of the influence of site accessibility on attracting passenger flow and calibrates the quantitative weight coefficients of the location conditions on the transfer passenger flow based on the attraction range constraints determined in the previous sections.In order to determine the new line station category,for new stations,this paper constructs the mapping relationship between the station category and the land use index,and by extracting POI(Point of interest)data and AFC(Automatic fare collection system)data,calculates the corresponding correction coefficient to correct the passenger flow of the existing station of the same category as the predicted value of the new line station.Finally,in order to verify the applicability and accuracy of this model,the middle section of Beijing Metro Line 14(Beijing South Railway Station-Jintai Station)was used as an example to predict passenger flow for the newly opened stations under the new line.By analyzing the historical data,it is found that the prediction of natural change passenger flow is more accurate when using the ARIMA method.The absolute percentage of the average error of subsequent predictions based on this method is 3.87%,indicating that the ARIMA model can better capture the unconventional changes in passenger flow after the new line access.The ARIMA model also verifies that the proposed passenger flow prediction method considering the land use and accessibility of the new rail transit station under the constraints of the attraction range has high accuracy,and provides a new idea for the passenger flow prediction in the establishment and feasibility study of the new rail transit line.