Fuzzy-theory-based pedestrian dynamics models for studying the waiting passenger distribution at the subway platform

Abstract

The behavior of waiting area choice of passengers at the subway platform can directly affect the travel efficiency, thereby affecting the service level of the station. Combining fuzzy logic theory and cellular automata model/social force model, a waiting area choice model for passengers at the subway platform is proposed, which can be used to predict the distribution of passengers. The distance from passengers to waiting areas, the number of passengers in the waiting area, and the length of the remaining waiting area are selected as the main influence factors after the analysis of field investigation, which are defined as the input variables of the fuzzy logic system. The choice probability of passengers to each waiting area is defined as the output variable. The waiting area with the highest probability is taken as the destination of passengers, thereby the static field value in the cellular automata model and the desired direction in the social force model can be determined. The total error of the established method can be as low as 10.13%, which shows a relatively good prediction effect. The effects of passenger attributes, stair attributes, and passable area size on the waiting area choice behavior are simulated. The results show that the influence degree of passenger companion behavior on passenger distribution depends much on the proportion of passengers in the company. The body radius and the desired speed of passengers have a great impact on the distribution balance of waiting passengers based on the analysis of the Gini coefficient and random forest method. The interval of multiple stairs should be the same as far as possible. The setting of a passable area is conducive to the flow of passengers during peak hours. The number of passengers is most pronounced in the distribution balance of passengers by comparing five different influence factors. Guidance strategies can be released through the LED display screen on the basis of the predicted distribution information. This paper provides a new idea for modeling the waiting area choice behavior of passengers, which lays a theoretical basis for the design and transformation of platform facilities.