| Urban rail transit has become an important part of public transportation because of its advantages such as high speed and high security.With the rapid development of rail transit in domestic cities in recent years,the network structure is becoming more and more complex,and the choice of routes on rail transit is increasing,and the passengers’ flow distribution of the network changes with it.The domestic rail transit generally adopts the No-Barrier transfer mode and the Automatic Fare Collection System to realize the ticketing and checking activities of passengers.The system only records the time,place and amount of passengers entering and leaving the station,and does not record any information of the transfer.We don’t know the actual choice of passengers,as well as the transfer station when passengers cross the line transfer.For the rail transit with complex network structure,there are usually many optional routes between the origin and the destination.The key to analyze the passengers’ flow distribution of the rail transit network is to determine the selection ratio of each alternative route.The study of passengers’ flow distribution is the basic work of rail transit operators’ new line planning,business plan formulation and the ticket fare clearing.The direction of traditional research mainly refers to the equilibrium and disequilibrium theory in the field of road traffic flow,analyzes the travel behavior of passengers through artificial investigation,and quantifies the factors that affect the choice of passengers’ path to study the passengers’ flow distribution of rail transit.Under the background of the complex network structure of rail transit and the dynamic adjustment of operation plan,the applicability of the traditional method is reduced.In recent years,it has become a new trend to study the passengers’ flow distribution of rail transit with the help of passengers’ travel data of traffic card.In this paper,based on Gaussian mixture model and the real passengers’ data of traffic card,a research method of passengers’ flow distribution of rail transit under the No-Barrier transfer mode is put forward.This paper mainly carries on the following research work.Firstly,the characteristics of rail transit network are analyzed,and the OD optional route of rail transit is divided into single-route OD and multi-route OD.The unbalanced distribution of passengers’ flow is explained in terms of time and space dimension.Based on this,the peak of passenger flow of rail transit is divided.Secondly,based on Gaussian mixture model,taking passengers’ travel time as observable variable and passengers’ choosing route as non-observable variable,we can construct the multiroute OD model of passengers’ flow distribution.The travel time is divided into three stages by analyzing the constituent factors of travel time and the factors that affect the choice of passengers’ routes.Using single-route OD’ data,the elements of travel time of optional routes are estimated as the initial parameters of Gaussian mixture model.Finally,through the real passengers’ data of Shanghai rail transit,the feasibility and accuracy of the passengers’ flow distribution model proposed in this paper are verified,and the availability of the method of segmentation of travel time as the parameter initialization method of the Gaussian mixture model are explained. |
