Tracing and prediction analysis of an urban pipeline leakage accident based on the catastrophe DBN model

With trends in urbanization, many underground pipelines have been constructed, forming a huge spider web-like network, and accidents in such systems can lead to disastrous consequences, as observed in the Qingdao and Kaohsiung gas explosion accidents. For such a complex system, questions related to identifying the accident causes and predicting the evolving trends represent research challenges. Such accidents are often catastrophic and dynamic, and traditional methods are only appropriate for static accident analyses. Based on the above problems, an accident tracing and prediction method based on the bow-tie model and the catastrophe theory dynamic Bayesian network (DBN) model (C-DBN) is proposed in this paper. First, a fishbone diagram (FD) is used to analyze all possible causes of the accident, and the bow-tie and Bayesian network (BN) models are established according to the logical relationship. The conditional probability tables (CPTs) of the BN are obtained using the transformation rule and catastrophe theory. Dynamic analyses of the accident causes and development trends can be conducted using the powerful reasoning ability of the DBN, which can supply guidance for emergency response and rescue. Finally, the Kaohsiung gas explosion accident is used as an example to verify the feasibility of the method. The results show that the analysis results are highly consistent with the real scenario.