A Prediction Model and Factor Importance Analysis of Multiple Measuring Points for Concrete Face Rockfill Dam during the Operation Period

Dam settlement monitoring is a crucial project in the safety management of concrete face rockfill dams (CFRD) over their whole life cycle. With development an automatic system, large amount data was collected. To precisely predict structural health dams, combined multiple points (MMP) model and machine learning has been developed. In this paper, based on physical factors CFRD, we comprehensively analyzed influence water level load transfer, rheology soil properties during impoundment operation period. Then, established space-time distribution CFRD its period under factors. An extreme gradient boosting (XGBoost) used for fitting prediction, evaluated using various performance indicators. The results show that spatial parameters such as upper filling height, thickness, panel-point distance material correlate to deformation characteristics dam. Taking Liyuan example, new MMP full-section with higher accuracy, which certain application popularization value related projects. evaluate contribution components model, SHapley Additive explanation (SHAP) methods are importance selected factors, reasonability these verified.