Coordinated Passenger Flow Control Strategy Of Metro Line Based On OD Passenger Flow Prediction

Urban rail transport is becoming the primary mode of travel for urban residents due to its high capacity,comfort and reliability.However,during peak hours,its capacity is still unable to meet the travel demand of residents,and operators need to formulate corresponding flow restriction strategies based on experience.In this paper,we analyze the spatial and temporal characteristics of passenger flow based on the data obtained from the automatic fare collection(AFC)system,then make short-time OD passenger flow forecasts,allocate passenger demand based on the forecast results to different sections,and build a linear programming model based on the allocated passenger flow data,considering efficiency and the spatial and temporal balance of travel and solve it.The main research elements of this paper are as following:(1)Spatial and temporal passenger flow data processing,passenger flow feature analysis.The AFC data is used as the original data,and the passenger flow data is extracted and pre-processed to generate a component-time OD matrix.The passenger flow information is analyzed in multiple dimensions such as time,space and travel time.At the same time,the k-order adjacency matrix,shortest-circuit weight matrix and temporal similarity matrix are constructed in conjunction with the metro topological network.(2)Short time OD passenger flow prediction based on GCN-LSTM model.According to the statistics of different OD passenger flows at departure stations under15 min time interval,the OD pairs is ranked according to average OD flow and OD pairs with less flow are covered in the form of masks for OD matrix reconstruction.After normalizing the temporal and spatial data,a GCN-LSTM model is built to train the model at 15-min time granularity and perform short-time OD passenger flow prediction.(3)Construct a generalized travel cost to search for and allocate passenger flows between OD pairs of valid paths.The calculation of the generalized travel cost of travel paths is proposed,whereby the depth-first algorithm of the graph is used to search for the set of valid paths,a stochastic Logit model is built and solved using the successive averaging method to determine the proportion of passenger flows allocated to different paths.(4)Propose collaborative traffic flow restriction strategy under vehicle flow and passenger flow matching.Based on the above OD prediction and passenger flow allocation results,a collaborative flow restriction model is established by matching passenger flow with traffic flow during peak hours,considering travel efficiency,fairness and balance,and the collaborative flow restriction strategy is derived by solving using the simulated annealing algorithm.(5)The method proposed in this paper is validated by taking Beijing urban rail transit as the research object.