Research On Multi-path Passenger Flow Allocation Algorithm In Rail Transit Network

Urban rail transit is an important support for residents’ travel in big cities in China.With the continuous increase of network mileage,network operation conditions are gradually mature.Under the condition of single route,the rail transit travel mode is simple and the travel path is unique.Under the condition of network,the accessibility,connectivity and flexibility of each rail station are enhanced,and the travelers can choose more travel routes.These routes have great differences in congestion,transfer convenience and time reliability.Combined with the attributes of passenger group and rail transit travel path,this paper studies the behavior of residents’ travel path choice under the condition of network,and studies the OD allocation prediction of rail transit passenger flow.Firstly,the evolution of urban rail network and the spatial and temporal distribution characteristics of passenger flow are studied.This paper discusses the causes of urban rail transit multi-path and the differences between paths.Combined with AFC big data,the temporal and spatial distribution of passenger flow between different od point pairs is calculated,and the temporal,daily,weekly and monthly variation rules of od are described.In addition,the distribution characteristics of rail transit passenger flow in line section,the differences of different stations in peak hours,and the macro distribution characteristics in group and line scale are studied.Secondly,the travel behavior of rail transit passengers is discussed from three dimensions of travel intensity,travel time and travel space.Based on the travel characteristic indexes such as travel days,average daily travel times,average travel time and first travel time,passengers are divided into four travel categories by using fuzzy cmeans clustering method.Through the analysis of the travel index data of four types of passengers and the characteristics of passengers,the paper analyzes the characteristics of the four types of passengers According to the distribution of time and space,the attributes of all kinds of passengers are analyzed,and the key factors affecting the travel behavior of rail transit passengers are extracted.Thirdly,the topology of the rail network structure is analyzed,and the depth first algorithm based on graph is used to search the effective path.Considering the perceived differences of different types of passengers in travel time,transfer reliability and congestion,a generalized cost function based on passenger differences is set to quantify the travel utility of different passengers.By constructing the probability model of passengers’ choice of route,the probability of different types of passengers’ choice of route between OD pairs is estimated,and the parameters of the model are solved by combining with multiple linear regression estimation method.Based on this,a multi-path stochastic probability passenger flow assignment algorithm is designed to calculate the line section flow.Finally,the OD of Chongqing rail transit passenger flow is taken as an example to analyze the passenger flow distribution,and the results of different passengers’ differentiated choice on the path are obtained.When there are multi-path in rail transit network,commuters tend to choose the path with shorter running time;when the running time difference between multi-path is less than 3 min,passengers prefer the path with less transfer times;in addition,non commuters tend to choose the path with lower congestion,and long-distance non commuters tend to choose the path with less transfer times.The above factors lead to the unbalanced distribution of OD passenger flow in the multi-path area.The stochastic probability distribution model can more accurately describe this phenomenon,and accurately calculate the actual passenger flow undertaken by the rail transit section,which can provide a reference for the operation management.