Research On Crowd Simulation Model Based On Multi-Agent

With the economic development and social progress, our country urbanization advancement speeds up unceasingly and various large buildings are becoming more and more. All kinds of sudden accidents frequently occur. Setting evacuation plans is especially important when a lot of people gather at some place in case of any emergent situations. As the process of human evacuation poses a threat to the public safety, it would brought about the serious result if is not handled properly. So, in recent years, the study of evacuation has become one of the focuses of the safety study.Crowd behavior is a very complex phenomenon, which not only depends on psychological factors associated with its own, but also depends on many other factors. Thus, it is very difficult to describe crowd behavior with a set of mathematical formulas. As the evacuation exercise need a large number of personnel and costly, therefore, the main feasible tool can be used to study evacuation is computer simulation.This dissertation has brought the perspectives of psychology and sociology about human behavior in emergencies into computational models for egress analysis. Diverse human behavior has been incorporated into a Multi-Agent Simulation System (MASS) for Egress analysis. MASS adopts a multi-agent based simulation paradigm to model individual behavior. Each agent has equipped with sensors, brains and actuators. Individual behavior is simulated through modeling of sensing, decision-making, behavior selection and motor control. Social behavior, such as Competitive, queuing, herding, and leader-following behaviors, is simulated through modeling of individual behavior and interactions among individuals. Simulation results are displayed 3D visual images by using visualization. MASS is a modular-based computational framework, which has good extensibility and can easily incorporate new behavior types.In this model, point-test algorithms, ray-tracing algorithms and decision-trees are incorporated into MASS to simulate the sensing, decision-making, behavior selection, and motor control of agent. A Grid Method is utilized to perform collision detection among large number of agents with an O(N) time complexity. We adopt modified particle swarm optimization (PSO) algorithm as the path finding algorithm to compute the direction and position of an agent. Based on different environment situation, PSO choose a suitable Objective Function to more complex human behavior. In the object function of PSO, a number of computing factors have been eliminated, so this method is very efficient without losing the authenticity of the premise.