Multi-agent Network Modeling for Rapid Response Against Public Emergency in Communities
Open Access
- Author:
- Shi, Xiaoru
- Area of Honors:
- Industrial Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Hui Yang, Thesis Supervisor
Catherine Harmonosky, Thesis Honors Advisor - Keywords:
- Community network model
decision-making
evacuation simulation
multi-agent modeling
pedestrian flow.
Community Network Model
Decision-Making
Rapid Response Simulation
Multi-Agent Modeling
Pedestrian Flow - Abstract:
- Heavily populated communities such as cities, universities, and townships are at heightened threat when facing public emergencies, susceptible to significant financial losses and casualties due to their intricate landscapes and high population mobility. The critical need for robust and effective emergency response policies is acknowledged by community administrations. However, there exists a lack of comprehensive analysis of community-wide response processes. Furthermore, little progress has been achieved in establishing a community pedestrian network model to scrutinize response procedures for multiple concurrent and diverse emergencies. Thus, there is an urgent need to develop a simulation model that reflects community layout, and pedestrian flow and accurately represents pedestrian behavior in response to diverse emergencies. This paper unveils a multi-agent community network model to simulate and assess pedestrian response processes in emergency scenarios. First, we develop a community network model that represents different complex real-world communities with flow networks, utilizing publicly accessible map data. Second, we propose human and hazard agents to model pedestrians with diverse pedestrian behaviors and emergencies with distinct characteristics. Lastly, simulation experiments are conducted to assess the correlation between the panic level and the efficacy of response procedures. The designed multi-agent community network model is corroborated through experiments involving simulated emergency responses staged at a large-scale campus under uncertainties from pedestrian flow and hazard occurrences. Experimental results validate the ability of the proposed approach to represent pedestrian flows on complex campuses and provide sufficient environments for response process evaluations. The outcome analysis of simulation experiments provides valid evidence that matches the nuances of pedestrian behavior proposed in the model in response to diverse emergencies in complex communities. This advancement offers significant potential to aid community administrators and researchers in developing and testing resilient emergency response strategies in case of public emergencies.