The Passenger's Travel Network And Its Dynamics Based On The Mass Data Of Intelligent Card

Public transport system plays an important role in modern society.Traffic problems attract great attentions whether in practice or academia.With the continuous popularization of public transport and increment of passenger trips,the problems,such as poor cooperation between bus system and subway system,traffic congestion and low efficiency of urban public transport system low become more and more serious.How to enhance the transport efficiency of urban public transport system has become a hot topic in the field of transportation.In the era of the big data,we can not only study topology and robustness of the public transport network,but also can mine passengers' behavior rules.It's of great practical significance for the promotion of the efficient cooperation between the public transportation and the subway and the comprehensive transportation capacity of public transport.First of all,in this paper,we analyze the topology and robustness of the bus and subway interdependent network.Then we divide subway passengers' travel network into several communities by an improved Particle Swarm Optimization algorithm based on the Physarum-inspired model.Finally,we predict the subway passengers' flow by a new regression method based non negative matrix factorization and auto regression.The main contents are as follows:(1)We construct the bus and subway interdependent network,and make a comparative analysis between the topology and robustness of interdependent network and sub-networks: Based on two different modeling methods(Space L,Space P),the interdependent station network and transfer network are constructed.Meanwhile,the topology of two kinds of complex networks and their corresponding sub-networks are analyzed and compared.Besides,we also make a comparative analysis of robustness index(such as Relative size of the giant component,Average path length,Diameter,Performance parameters)of interdependent network and its sub-networks under different attacking mode.In this paper,we present the empirical investigation results for the urban subway networks through the network data of a Southwest city of China.The result shows that he bus and subway interdependent network and its sub-networks are scale-free small world networks.(2)We complete the community partition of the passenger travel network: we get the initial solution of obtain the objective function,namely the network modularity function,by Physarum-inspired model,and then dividing the passenger travel network into communities by combined with PSO algorithm.Based on the passenger travel network,we compare P-PSO with the greedy discrete particle swarm algorithm(GDPSO).The results show that: in the community mining of weighted network,the modified particle swarm algorithm based on Physarum-inspired model has been significantly improved in terms of the feasibility of the solution.(3)We predict the subway passengers' OD matrix with the help of NMF-AR model: Firstly,we get feature vectors by NMF algorithm.Then,based on the factorization of NMF,the coefficient matrix is established to predict the passenger flow.Meanwhile,we make a comparative analysis of among these method(such as KNN,C4.5,NB,RF)based on the subway passenger flow data of the city.The result shows that the prediction accuracy of the algorithm is significantly improved.