Research On Estimation Method Of Stranded Passenger Flow Under Metro Train Interruption

As a huge and complex system,once the train operation is interrupted in the subway system under network operation,it will seriously affect the normal travel of passengers,cause passenger flow retention and easily lead to safety accidents.When certain conditions are met,the subway management department will initiate emergency connection of ground buses for emergency response.However,in the actual operation process,due to the lack of precise grasp of the stranded passenger flow under interruption conditions,the emergency department often finds it difficult to ensure the rationalization of emergency resources The configuration is also not conducive to ensuring the maximum degree of passenger travel.Therefore,a reasonable estimation of the number of stranded passenger flows and the distribution of their destinations under interruption is of great significance in emergency aid decision-making.Based on specific interruption scenarios,this paper analyzes and defines the stranded passenger flow under the interruption of subway train operation,and divides it into two types of platform stranded passenger flow and train stranded passenger flow according to the location at the time of the interruption,and proposes a subway inbound passenger flow prediction model Based on the distribution prediction model of passenger flow OD(Origin-Destination,passenger travel origin and destination),based on the analysis of a large number of card swiping data,two methods for estimating the stranded passenger flow are respectively proposed,and finally a case study is carried out.The specific research content is as follows:(1)Construction of subway train interruption scenarios and analysis of stranded passenger flow.A typical subway train operation interruption scene is constructed,and the corresponding train operation adjustment analysis is carried out.On this basis,the scope of the stranded passenger flow is defined,and the stranded passenger flow is divided into two types of platform stranded passenger flow and train stranded passenger flow,and the research object is clarified.(2)Short-term inbound passenger flow prediction and OD distribution prediction model construction.Based on Graph Convolutional Neural Network(GCN)and Gated Recurrent Unit(GRU),a GCN-GRU model for subway short-term passenger flow prediction is proposed and designed,using real subway passenger flow Historical data,training and evaluation of the model,verified its superiority in short-term passenger flow prediction.The passenger flow OD distribution prediction model is further designed,and the historical OD distribution ratio matrix is continuously updated through the exponential smoothing method to obtain the OD distribution ratio matrix of the current period,so as to predict the OD distribution of passenger flow in the current period.The construction of the two models provides a data basis for the following passenger flow estimation.(3)The design of the estimation method for the stranded passenger flow at the platform.Analyze the formation process of the stranded passenger flow on the platform from the time point of view,establish the passenger travel time chain model,and estimate the relevant time parameters based on the statistical analysis of a large number o f card swiping data,discuss the probability of the passengers not getting on the train in different time periods,and design Monte Carlo simulation experiment is to estimate the passengers who have not boarded the train,and carry out statistical analysis on the experimental results,so as to obtain the method of estimating the passenger flow on the platform.(4)The design of the estimation method of the train detained passenger flow.A method for tracing the source of the passengers on the train is proposed to estimate the passengers on the train,and then applied to different interruption scenarios to obtain the estimation method of the train’s stranded passenger flow.The specific content includes:estimating the passenger travel time chain to obtain the theoretical travel time of the passenger;calculating the effective path set of each OD pair of the subway network,and estimating the route selection ratio of the passenger by comparing the actual travel time of the passenger with the theoretical travel time;Based on the train timetable and driving information,trace the origin of the OD pairs of passengers on the effective path through a certain train section,predict the inbound passenger flow and OD distribution of these ODs to the starting point,and carry out the route selection ratio.Estimate,thereby judging the passenger boarding situation between each OD pair when the train stops,so as to complete the estimation of the passengers on the train.(5)Analysis of calculation examples.Based on the local subway network of Shanghai in 2015 and a large amount of credit card data,Simulate the interruption of metro train operation,and the stranded passenger flow estimation method proposed is used to estimate the stranded passenger flow to verify the feasibility of the method.