Modeling Emergency Events Chain Based On Bayesian Networks

Recent years, emergency events have brought huge lost to the human society, which calls the eyes of government and academic circles. Emergency event is the object of emergency management. Understanding and predicting the evolving process of events can improve the validity of emergency management. However the related studies have the following shortcomings, on the one hand, the recent studies focus on the single emergency event, lacking the study from the angle of system, which can not meet the needs of comprehensive emergency management. On the other hand, the related studies have not provided a universal modeling method for all kinds of events.Concerning the shortcomings in related studies, based on the theory of hazard system, disaster chain and emergency event chain, this paper firstly views emergency event as a system, and the basic elements of events is proposed, including elements of input, state and output. Then the relationship between events is analyzed from the prospective of coupling theory, and the emergency event coupling degree model is proposed. After analyzing the emergency event from the system view, this paper proposes a single event model based on Bayesian Networks, including the selection of nodes, the study of network structure and network parameters, then the emergency event chain model is proposed by combing the events Bayesian Networks, the method of combing networks is proposed, the structure of event chain network is analyzed, and the inference issues are also discussed. At the end of this paper, typhoon event happening at Taiwan area is incorporated as an example to demonstrate the method proposed in this paper.The work in this paper is of significance for emergency management. Understanding the mechanism of all kinds of events from system view can help treat events in a integrate way, which is useful for comprehensive emergency management. The modeling method proposed in this paper can model all kinds of events and event chain, so that predict the evolving process of events and events chain, which can help emergency agencies make well preparations, reducing the lost of human society brought by events.