Identification of Parameters fora Social Force Model in Evacuation Simulation Using Evolutionary Computation

Abstract—Preparations for possible disasters should be made to minimize the damage caused by events such as fires and earthquakes. Several kinds of disaster simulations have been developed for this reason. Many evacuation simulation methods are based on multi-agent systems. These are considered to be useful and powerful methods of dealing with complex systems containing several elements, and have been proposed for safely guiding evacuees. The social force model is one approach that considers the influences and interactions between evacuees. When aiming to improve the precision of social force model simulations, the model parameters should be appropriately determined. In this paper, a method for determining the parameters used in the social force model is proposed by applying an evolutionary computing method to collect evacuee flow data.