Research On Departure Time Choice Behaviors Of Urban Rail Transit Passengers During Morning Peak Hours

With the rapid development of China's social economy,urban rail transit plays an important role in the transportation of passengers and alleviating road traffic pressure.However,with the gradual growth of passenger flow volume,there is a mismatch between passengers travel demand and transport capacity of stations during peak hours,which leads to serious congestion in stations,lines and networks.This issue not only brings a lot of difficulties in the operation and management of the subway,but also leads to the potential security problems.In order to release the crowding during peak hours and control the passengers' travel demand,this paper analyzes the departure time choice behaviors of passengers during morning peak hours under the condition of network operation with heavy passenger flow volume.The main work in this paper can be summarized as follows:(1)Travel preferences of urban rail transit passengers during morning peak hours are studied based on the analysis of temporal and spatial departure time distribution of urban rail transit.The process of departure time choice is analyzed and the main factors influencing the departure time choice are summarized.A method combining stated preference(SP)and revealed preference(RP)is proposed to obtain the data of departure time choice.First of all,the key attributes are analyzed and selected using the relative to an identified distribution unit(RIDIT)method.Revealed preference data is utilized to design the attributes of stated choice(SC)experiments in order to enhance the rationality of stated choice experiment scenarios.Finally,the D-efficient is employed to optimize the attributes allocation of stated choice experiments,which improves the model parameters estimation and the overall fitness.(2)Departure time choice behaviors of urban rail transit passengers are analyzed and predicted based on existing models.Firstly,the mixed logit(ML)is employed to analyze the characteristics of departure time choice,and maximum likelihood estimation(MLE)is used to estimate the parameters.Then,the impact of attributes is analyzed through elasticity analysis and preference heterogeneity of passengers is studied,which provides a theoretical basis for making passenger travel demand control strategy.Then,Bayesian network(BN)is employed to predict the departure time choice behaviors of passengers during morning peak hours.The K2 algorithm and Bayesian estimation are used for structure learning and parameter learning separately.And network inference is achieved through junction tree algorithm,which verified the effectiveness of proposed Bayesian network further.Finally,receiver operating characteristic(ROC)curve is utilized to compare the prediction accuracy of mixed logit and Bayesian network,and Bayesian network displayed higher prediction accuracy in the inference of departure time choice behaviors.(3)A case study of models application for subway passenger travel demand control is conducted,which takes Batong line of Beijing urban rail transit as an example.Firstly,distributions of inbound passenger volume from 6:00-9:00 in all stations are analyzed,and the strategy of carrying out fare discount before morning peak hours in Tuqiao,Guoyuan,Shuangqiao and Communication University Station is proposed.Then,a simulation-based method is used to estimate the train loading factor,and the ratio of high train loading factor and distributive entropy of Euclidian distance are employed to evaluate the passenger travel demand control strategy.The results show that the proposed travel demand control strategy has changed the departure time choices of some passengers,which releases the overall crowding of the line.