Abstract: The scenario modeling of unconventional sudden disasters and accidents has abundant information resources, but it is faced with several problems, e.g. numerous sources, obvious heterogeneity and insufficient microanalysis. To make up for these deficiencies, the knowledge element representation method was used in this paper to describe sudden disasters and accidents from the aspects of scenario state, emergency rescue activities, disaster inducing environment and disaster bearing body. Then, based on the uncertainty and self-organization characteristics of disaster and accident evolution, a dynamic scenario model based on dynamic Bayesian network for disasters and accidents was constructed. Finally, the scientificalness and effectiveness of the dynamic scenario model based on knowledge element and Bayesian network for unconventional sudden disasters and accidents were verified by conducting a case study on the "11·22" Qingdao leakage and explosion accident of oil pipeline. It is shown that the scenario evolution path presented in the dynamic scenario model is basically the same as the evolution sequence of the actual pipeline leakage accident. The research results can provide a reference for the prevention of unconventional sudden disasters and accidents.