The current study proposes an evacuation optimization model (IEO model) for mass emergency evacuation, which aims to minimize the average completion time, avoid traffic congestion, optimize the network utilization rate, and balance the exit loads. Pedestrians are divided into free evacuation pedestrians and organized evacuation pedestrians. The K-means algorithm and improved parameter adaptive DBSCAN algorithm are adopted to divide organized evacuation pedestrians into appropriate groups. Combined with an improved ant colony algorithm and the collision avoidance strategy, route scheduling for organized evacuation groups is carried out. A representative case is employed to benchmark the performance of the proposed model. The universality of the proposed model targeting different failure time of key exit and evacuation scale are analyzed. The developed model shows obvious advantages in mass evacuation compared with traditional statistical methods. 

Jingwie Ji

Papers

Research and realization of parallel algorithms for large scale crowd evacuation in emergency