Passenger Flow Analysis Of Urban Rail Transit Under Network Operation And Passenger Flow Forecast Of New Lines

With the accelerated construction of urban rail transit and the improvement of network level,the uncertainty of passenger flow changes of existing lines and new lines after the opening of new lines has created a greater challenge for the network operation and development of urban rail transit.In order to solve the above problems and provide reliable passenger flow data to operation managers at the early stage of new line opening,this paper,based on the idea of traditional four-stage method and summarizing the current situation of related researches at home and abroad,researches the theories and methods of passenger entrance and exit flow prediction,passenger flow distribution prediction and passenger flow assignment respectively for the specific scenario of new lines opening of urban rail transit.The research work in this paper is as follows:(1)This paper completes the analysis of passenger flow characteristics under networked operation conditions,in both regular operation and before and after the opening of new lines of different categories scenarios.The analysis perspective covers time dimensions such as week,day and hour and spatial dimensions such as network,lines and stations;the analysis indexes involve entrance flow,exit flow,OD,average distance,etc.Through the analysis,the conclusion that the changes in passenger flow are closely related to the land properties around the stations is obtained,which provides the theoretical basis for passenger flow prediction.(2)Based on the BP neural network,the improved PSO-BP algorithm was used to achieve the prediction of the entry and exit volumes of new and existing stations at the early stage of new line opening,using the area and affiliation coefficients obtained through the Baidu Map API as input data,and the accuracy of its prediction results was verified by test samples.(3)Based on the study of the existing gravity model,a tri-constrained gravity model considering station types is proposed to achieve passenger flow distribution prediction.In the process of determining stations’ types,stations are classified into six types by k-means++ clustering algorithm using the land use data and passenger flow data around the station.In determining the impedance function of the gravity model,the psychology that passengers will magnify the interchange time is fully considered,and a comprehensive impedance calculation method that considers passenger walking time,travel time and transfer time is proposed.In addition,this paper also proposes the separated recursive Dijkstra algorithm to improve the calculation speed of k-path in the passenger flow allocation process.(4)Meanwhile,passenger flow,passenger flow distribution and passenger flow assignment are predicted for new and existing stations of C metro in China under the real scenario of new lines opening,and the prediction results are compared with the classical BP algorithm and double-constrained gravity model.The comparison results confirm that the passenger flow prediction model and algorithm proposed in this paper have certain improvement in accuracy compared with the traditional prediction methods,and show the feasibility and effectiveness of new lines opening passenger flow prediction method proposed in this paper in the practical application process.There are 57 figures,37 tables and 88 references in this paper.