A High-Robust Displacement Prediction Model for Super-High Arch Dams Integrating Wavelet De-Noising and Improved Random Forest

Abstract

We present a novel deformation prediction model for super-high arch dams based on the prototype monitoring displacement field. The noise reduction processing of the monitoring data is conducted by a wavelet technique. The performance-improved random forest intelligent regression approach is then established for constructing the arch dam deformation statistical models, whose hyper-parameters are intelligently optimized in terms of the improved salp swarm algorithm. In total, three enhancement strategies are developed into the standard salp swarm algorithm to improve the global searching ability and the phenomenon of convergence precocious, including the elite opposition-based learning strategy, the difference strategy, and the Gaussian mutation strategy. A prediction example for super-high arch dams is presented to confirm the feasibility and applicability of the prediction model based on five evaluation criteria. The prediction results show that the proposed model is superior to other standard models, and exhibits high-prediction accuracy and excellent generalization performance. The stability of the proposed prediction model is investigated by artificially introducing noise strategies, which demonstrates the high-robust prediction features and provides a promising tool for predicting carbon emissions, epidemics, and so forth.