Passenger distribution modelling at the subway platform based on ant colony optimization algorithm

In the subway platform, not all passengers distribute randomly but gather in the waiting areas, especially when a train is coming. During emergency evacuations, passengers' initial distribution may play a significant role in affecting the escape efficiency. In this paper, a passenger distribution modelling method is proposed to predict such waiting area choice processes based on ant colony optimization (ACO) algorithm, which is really a complicated job due to many influence factors. The model considers the distance to the target waiting area, the length of queues, the physical length of waiting areas and the train schedule as four main influence factors. Specially, a modification of the passenger's impatience factor in the famous social force model (SFM), better reflecting the change of psychological states with an arrival of a train, is presented. The field data collected at the Xuanwumen subway platform is utilized for the model calibration and validation. The ultimate simulation results demonstrate that passenger distributions based on ACO algorithm basically can reflect the field distribution and also the dynamic characteristics of waiting area choice processes. Impacts of passenger distributions on evacuation dynamics under fires are further studied based on the software FDS+Evac. The results indicate that passenger distribution does has little impact on evacuation efficiency when fires are not very large, while the evacuation will be affected significantly by passenger distributions once fires are large enough. This further indicates the necessity of studying the passenger distribution at the subway platform especially under emergencies.