| As the service interface for the interaction between the urban rail transit (URT) system and passengers, the crowding management for passenger flow in peak-hour URT platform is the bottleneck of improving the URT network operation efficiency. The crowding in the platform is the result of the complex passenger flow's spatial-temporal evolution by self-organization and hetero-organization. Under crowded condition, variable passengers' crowding perception for platform space environment, static environments such as facility layout, as well as dynamic environments such as complex passenger flow affects the behavior significantly, leading to more obvious difference of passenger flow's dynamic characteristic on the platform compared with other facilities in the URT system. Therefore, it is meaningful to study the influence factors and characteristic of the passengers' crowding perception on the URT platform comprehensively, and develop the passenger's behavior model on the URT platform by considering passengers' psychological and dynamic behavior characteristics under the crowding condition. These works is helpful for studying local interactions under a complex environment resulting in the non-linear characteristics of emergences, and essential for the crowding management of platform in URT.This study focuses on developing dynamic passengers' behavior model based on the crowding perception in urban URT. The major work is fourfold:(1) Based on the analysis about the URT platform range, a behaviors observation experiment about typical platform passenger stations of Beijing Metro System has been conducted. A questionnaire about individual passenger crowding perception is developed based on video, which is the database for model development in this study. Meanwhile, a data collection experiment about the fundamental diagram of traffic flow is designed and conducted in this study. Based on the collected data, the velocity and density of passenger flow on the platform is analyzed, which supplies real-world data for the validation of simulation model. This study shows that the fundamental diagram of passengers flow on platform connecting channel or platform is consist with Greenshields Model, and the fitness is higher than0.74.(2) Based on the analysis of platform congestion and passengers'crowding perception, a passenger crowded perceptual preference experiment is designed to collect individual subjective preference data. Considering the discrete characteristic of collected individual subjective preference data, an ordered probit model is developed by considering the influencing factor, and the parameters in the model are estimated by Maximum Likelihood Estimate method. After that, marginal effects of18parameters, which affect the passenger crowding perception, are studied. It is found that per unit change of explanatory variables affects the passenger crowding perception significantly. The top9significant influence factors are as follow in a descending order:passenger volume, average speed, average density, gateway number, stairways and escalators, visual range, age, gate layout, the width of the platform. (3) According to the perception-decision-action-based analytical framework, the passengers' crowding perception on the platform is analyzed in cognitive science perspective, and the necessity of assumption of individual units and visual information are explained. According to the assumption of individual unit, the passenger is assumed as the agent with three kinds of abilities of environment perception, behavior decision, and action implementation. Considering balancing computational efficiency of large-scale agent simulation model against fineness of the model, a passenger adaptive agent model based on perception-decision-action is developed. According to the existing research in Highway Capacity Manual, the individual space of passenger is divided into two parts, which are private space and visual space, and the regression model of passenger individual space is developed based on empirical data. Based on the assumption of visual information, a concept of visual-information-based individual perceptive space is defined as the physical distance between the obstacles and nearest visual collision passengers. After that, a neuron based congestion perception function is developed, and the method for presenting visual-based passengers crowding perception is proposed. In this study, the passengers crowding perception is defined as the psychological collision distance between other passengers and obstacles in the visual range.(4) The key issue for developing passenger behavior model is to describe the direction and speed of individual passenger's movement accurately. According to the characteristic of individual cognition derives the decision, based on the information of crowding perception, heuristic rule and mathematic model about the direction and speed of passenger's movement on platform are developed. Two simulation scenarios are designed, which are pedestrian-walking-channel scenarios and complicated island platform scenarios. Based on that, a simulation model for passenger behavior on the platform is developed, and a reasonable testing program and specific measure is determined in order to validate the developed simulation model. The study shows that the individual crowding perception of visual-information-based passenger makes a specific influence on their behavior, and then it can better explains the mechanisms for passenger's crowding perception effecting passengers dynamic behavior, and the change trend is consistent with the analysis result on obviousness of factors affecting passengers' crowding perception. Meanwhile, in the passengers' crowding perception-based simulation model, the passengers'complex evacuating behavior is formed based on the local rules between individuals, and it can reproduce the macroscopic traffic characteristics of passenger flow under congestion condition reasonably, and the output data of the simulation model is consistent with the real-world data during the validation process. |
PDF Full Text Request