Abstract: There are many dynamic risk factors and uncertainties in the operation period of concrete dam. In this paper, a dynamic Bayesian network model is constructed to study the dynamic characteristics of dam breach probability by introducing time factors. In association with the Leaky Noisy-or gate extended model, the method of determining conditional probability is described. The time series curves of occurrence probability of dam breach against each risk factor are obtained for a practical project as a case study. The present model is proved to be prior to the static Bayesian network as the result is more rational. The research finding offers reference for the dynamic risk analysis and evaluation system construction of similar projects.

Key words: concrete dam, breach risk, dynamic Bayesian network, conditional probability, risk management