Image Processing Based Technology Of Extracting Microscopic Passenger Behavior Characteristics And Simulation Of Boarding And Alighting Passengers On The Subway Platform

ABSTRACT:Microscopic behavior of Urban Rail Transit passengers is the basis and precondition to explain the macroscopic laws of passenger flow, optimize the station design, as well as improve the operation of urban rail transit. For different purpose, passenger movements in the station can be divided in to several behavior categories such as in-off station, buying tickets, security check and boarding and alighting. Boarding and alighting time has some influences on stop time which may cause the fluctuation of the interval passing-capacity. At present, it is hard to study the behavior of passenger boarding and alighting because of the lack of data and theory. This study discusses the passenger boarding and alighting behavior as two aspects, data collection and model establishment, respectively.Traditional data collection techniques have some shortcomings such as the drawback of backward technic, slow collection and low accuracy, in that sense, these techniques of passengers cannot meet the requirements for passenger behavior researches. Image-processing data collecting technology, emerging and efficient, has been wildly used in road traffic but the application in the field of urban rail transit is still in the exploratory period. Based on the latest image processing technic, this paper independently developed a passenger boarding and alighting behavior characteristics collection system (PedTrace), to achieve the data extraction of speed, density and trajectory of pedestrian and provide the precise and actual data for behavior researches boarding and alighting.Because of the complexity and variety of passenger boarding and alighting behavior, traditional mathematical model cannot describe this behavior. With the social force model and Agent idea, this study used the methodology of microscopic simulation to establish the simulation system (PedSyms) to simulate the boarding and alighting behaviors which is separated into several parts like arriving and leaving, circumvention and waiting in line. In the model, a cell neighborhood based searching algorithm is used to reduce the complexity of the algorithm from N2to N, global path selection method is introduced to solve the travel path problem, collision avoidance method in the magnetic model is introduced to solve the pedestrian overlap problem, event-driven and agent-interactive technic is used to achieve the simulation model. Finally, taking Line2Xizhimen station boarding and alighting behavior as an empirical study, the speed and trace information was extracted by PedTrace and the speed data obey the lognormal distribution. Take the data as the input, the PedSyms simulated the boarding and alighting period to explore the behavior characteristics.