Abstract: Rainfall events are one of the main factors leading to tailings dam instability,and assessing the stability of tailings dams when rainfall events occur is critical to the safe operation of mines. Due to the variable mechanism of rainfall infiltration in the dam,the link between rainfall events and dam instability is uncertain,and to reduce this uncertainty,the Bayesian analysis was applied to predict the stability of tailings dams under different rainfall amounts. Based on the tailings dam stacking model test to obtain the physical parameters and test data of the dam body under different rainfall amounts and spatial dimensions,a Bayesian-based assessment model of tailings dam stability was established. Using the information gain to obtain the early warning value of each parameter and using it as an indicator to calculate the prior probability of the physical parameters, combined with the Leaky Noisy-or gat extension model to calculate the conditional probability,a Bayesian model is built with the help of GeNIe software,the probability of instability of the tailings dam at each spatial location under different rainfall conditions is obtained by Bayesian analysis. The model relates the experimental parameters to the uncertainty of the tailings dam destabilization mechanism,which further improves the accuracy of objective evaluation. The study results show that:① The greater the rainfall intensity,the greater the variation of the parameters of the tailings,the more obvious the interaction,the higher the probability of dam failure. ② Due to the spatially non-homogeneous nature of the tailings,the probability of tailings destabilization is greater at higher levels,and rainfall will further increase the risk of destabilization. The study results have some significance for the stability analysis of tailings dams.

Key words: tailings dam,stability analysis,Bayesian network,rainfall,GeNIe