A Combination Model for Displacement Interval Prediction of Concrete Dams Based on Residual Estimation

Accurate prediction and reasonable warning for dam displacement are important contents of safety monitoring. However, it is difficult to identify abnormal based on deterministic point results. In response, this paper proposes a model that integrates several strategies achieve high-precision interval displacement. Specifically, the realized in three stages. first stage, Extreme gradient boosting (XGBoost) constructed. second error sequence XGBoost generated by residual estimation method proposed paper, artificial neural network (ANN) constructed through maximum likelihood method. third noise composed training ANN carried out. Finally, results obtained above combined realize The performance verified monitoring data an actual concrete dam. show hybrid can not only better accuracy than single model, but also provide high quality