Primary Research on Urban Mass Panic Based on Computational Methods for Experiments

摘要

Base on the theory of computational methods of experiments and agent modeling technology, combining with sociology and psychology, the paper presents method of computational experiments for research of Urban Mass Panic (UMP). The method proposed a panic-information transmitting artificial system for computational experiments with agent modeling technology. The system planed two kinds of agents: social individual agents and panic alarming agents. Social individual agents are designed to simulate persons living common human society. These agents which contain the ability of panic information recognition, reflection and making different decision are simulated with some nonlinear methods, such as fuzzy neural network. Panic alarming agents are designed to simulate circumstance which engenders panic information or panic phenomenon. These agents collect information about panic and release information to social individual agents by different transmitting methods we can image. The paper also discusses how to carry through measurement to check the level of UMP by the information entropy theory dynamically.