An emergency evacuation system is a system that helps people in the space to evacuate safely and quickly from emergencies in the event of an emergency. Such systems are essential as the size of vessels becomes larger and more complex. However, current emergency evacuation systems play only a limited role. For example, evacuation route guidance through placement of real human resources or evacuation route such as direction of emergency exit point which is pointed in one direction only in one place. Relying on human subjective judgment in a dangerous situation can be quite dangerous, and emergency lights and escape routes that always point in the same direction are not able to deal flexibly with risk factors and can expose the public to danger. Furthermore, due to the nature of the ship structure, the initial response is important as the rescue time is delayed rather than the land accident.
Therefore, emergency evacuation systems should be more intelligent in increasingly complicated and larger structures, and should be able to quickly identify information on the surrounding situation and suggest an optimal evacuation route. In particular, it is not possible to exclude the possibility that dangerous elements may spread or become dangerous areas in the route where evacuees are passing.
Therefore, there is a need for a system that predicts and responds to the near future through sufficient modeling of risk factors. Among various risk factors, risk factors such as fire, smoke, and isolation can be sufficiently collected by using sensors or image processing devices. However, in the case of bottlenecks, it is essential to model the density of the population at the current node, the direction in which people at that location will evacuate, and whether the path of the selected path will accommodate the incoming population. Therefore, we propose a bottleneck modeling method and load balancing based on disaster situation in this paper. The proposed performance is verified by computer simulation.