Teaching Building Evacuation Modeling Based On Multi-agent And Path Optimization

The emergency evacuation in the campus is related to the personal and propertysafety of teachers and students. When emergency happens, a reasonable and effectiveevacuation plan is one of the most important measures to secure the safety.At present, most of the scenarios of teaching building evacuation system aredesigned into discrete models without considering the influence of such traits asgender, age and psychological factors of pedestrians. Because of this, the models cannot reflect the reality perfectly and may lead to potential dangers in practicalapplication.This thesis focuses on teaching building emergency evacuation modeling. It willanalyze the traits and behavior of evacuees and stimulate the decision-making processof path selection of pedestrians. The main work includes:1) Establishing a teaching building evacuation model based on Multi-Agent. Withthe Multi-Agent model, combined with the characteristics of evacuation under certaincircumstances of teaching building, we model the teaching building scenarios,analyze the traits and behavior of evacuees, abstract these individual factors intoagent-related properties to enable the individual differences among evacueespresented in designed model;2) Improve the basic particle swarm algorithm to guide pedestrian in a teachingbuilding evacuation. Use the idea of social force model to simulate the impact causedby surrounding environment on evacuees’ decisions during their walking process.Improve the best-individual path-select strategy in Basic Particle Swarm Algorithm tomake it more suitable for the heterogeneous agents under teaching building scenario;3) Design a simulation system upon the model and algorithm and then analyzethe simulation result to verify the system. A single-layered teaching building wasselected to conduct the simulation experiment in this paper. We analyze theexperimental data to prove the correctness and effectiveness of teaching buildingevacuation model. The model can more effectively simulate the entire evacuationprocess and the evacuation strategy base on improved particle swarm algorithm isbetter than that based on traditional shortest path algorithm to some extent. This thesis analyzes some individual factors that affect the entire evacuationprocess of pedestrians on the basis of Multi-Agent model，design and implement theMulti-Agent evacuation model based on continuous space under the teaching buildingscene, improve particle swarm optimization algorithm to guide the pedestrian in theevacuation process, and finally make evacuation experiments to prove the validity ofthe model and algorithm.