Analysis and Agent Modeling of Metro Passengers' Unsafe Behavior Based on Theory of Planned Behavior

Metro is a significant part of modern urban traffic system. But in recent years, metro accidents happen occasionally. Passengers' unsafe behavior is a significant cause. This research focuses on factors and influence path of unsafe behavior and tries to take passengers' attributes as the breakthrough point to propose appropriate suggestion to facilities improvement and evacuation plans. The theory of planned behavior is the fundamental theory of this research. Questionnaire design and its analysis are all based on five variables from theory of planned behavior and another four variables introduced from literature. Difference of personal attributes on unsafe behavioral intention and influence factors were analyzed. Structural equation model was used to analyze the influence path of factors. Passengers' personal attributes were transferred into speed parameters used in simulation. Hall and platform were used as two main simulation scene in Anylogic. Time and number of passengers in queue are recorded in whole progress of taking metro and evacuation. After all those analysis, respondents' opinions were shown on implementation intention, risk awareness and factors of unsafe behavior. Focusing group with specific attributes was pointed out. Influence path of factors was found by structural equation model. Suggestion to facilities improvement was proposed and verified. Appropriate evacuation measurements were proposed to increase the evacuation efficiency.