The exit selection behavior of pedestrians plays an important part in the process of evacuation. This paper proposes a cellular automata model based on fuzzy logic method for simulating the evacuation of pedestrians from a multiple-exit room. When pedestrians select the exit, the distance and density are adopted as two important input variables in the fuzzy logic method, and the probability of selecting each exit is defined as the output variable of fuzzy logic method. The output variable of fuzzy logic, exit width and herding behavior are combined to determine the target exit. The competitiveness of each pedestrian is calculated by logit model to solve the position conflicts among pedestrians. The validation of the model is demonstrated by comparing the simulation data with the real data. The effects of attributes of pedestrians, exits and obstacles on evacuation are studied in simulations. Results show that large public facilities should control the inflows of pedestrians, and the reasonable increase of the exit quantity and exit width are effective for improving the evacuation efficiency. In the design of buildings, obstacles need to be designed reasonably, which should not be too large or too small. At the same time, obstacles should be kept at a certain distance from the exit, so as to ease the exit congestion and improve the evacuation efficiency. This paper takes the advantages of fuzzy logic method to solve the exit selection problem, which can effectively integrate the robustness with physiological-based "perception-action" behavior, the experience knowledge of pedestrians and the perception information of the surrounding environment into the decision-making process. INDEX TERMS Cellular automata model, exit selection, fuzzy logic method, logit model, pedestrian evacuation I. INTRODUCTION

Alighting and boarding efficiency (A&Be) of passengers plays an important role in the formulation of subway timetable. This paper studies the impact of alighting area width on A&Be based on the improved social force model. This improved model adopts the fuzzy logic theory considering factors of train dwell time and passengers ahead to determine the desired speeds of passengers instead of setting fixed values. The t-test method verifies the validity of the model. The passenger movement dynamics characterized by the alighted or boarded passenger quantities, the desired speeds, and the actual speeds over time is then analyzed. The simulation results indicate when people keep to the civilized rules of alighting first, A&Be will be improved significantly with increasing the alighting area width. Otherwise, the arbitrary increase of alighting area width does not necessarily improve A&Be. The alighting area width during the actual design should be determined by the specific passenger volume and passenger civility. Whether the countermeasure of adding the fences to separate the alighting passengers from the boarding passengers is effective to improve the travel efficiency is finally explored through analyzing the simulation results. Our findings could provide theoretical supports for decision-making of waiting area division on the subway platform and passenger flow management in reality. INDEX TERMS Social force model, alighting area width, alighting and boarding efficiency, fuzzy logic theory.