A Computationally Efficient Method for Simulation-Based Evacuation Guidance Optimization

Effective evacuation guidance can help evacuees reach the exit as soon as possible in an emergency evacuation process to ensure their safety. However, due to the unstable mobility condition and crowd distribution, the guidance plan made in advance by comparing the distances to exits may not be the most effective, so it is necessary to provide real-time optimal guidance information in the plan. This study presents a computationally efficient evacuation plan optimization method that combines CTM (Cell Transmission Model)-based simulation and DRF (Directed Rooted Forest)-based planning. In this method, the simulation module predicts evacuation dynamics at a computational cost that does not grow with crowd size, while the planning module takes advantage of the simulation feedback to optimize the evacuation plan efficiently. The high efficiency is achieved by a DRF structure suitable for representing the evacuation plan and a node assignment optimization algorithm highly coupled with the CTM simulation. In addition, the feedback from the simulation is employed to evaluate the performance of new solutions, making the search more directional and accelerating the algorithm convergence. The proposed method may serve as the foundation for developing real-time evacuation guidance plans for large-scale crowded buildings.