An Explainable Probabilistic Model for Health Monitoring of Concrete Dam via Optimized Sparse Bayesian Learning and Sensitivity Analysis

Machine learning has become increasingly popular for modeling dam behavior due to its ability capture complex relationships between input parameters and responses. However, the use of sophisticated machine methods monitoring behaviors making decisions is often hindered by model uncertainty a lack interpretability. This paper introduces novel health monitoring, focused on radial displacement seepage, using optimized sparse Bayesian sensitivity analysis. The hyperparameters are an intelligent optimization method integrating multi-population Rao algorithm blocked cross-validation, while analysis employed calculate relative importance variables better understanding dam’s state. effectiveness proposed verified long-term data prototype concrete arch dam. results confirm that provides satisfactory performance both point predictions interval structural obtaining effective explainability.