Centroid deformation-based nonlinear safety monitoring model for arch dam performance evaluation

Abstract

The purpose of this study is evaluating the arch dam performance through modelling on the defined centroid deformation with the measured data. The space-time distribution characteristics of the arch dam deformation are firstly identified through the comparative analysis of typical arch dams with height of 200–300 m in China. Subsequently, the geometry centroid of the deflection curve of an arch or a cantilever, composed of multiple monitoring points, is defined to indicate the global structural behavior of the arch dams. In order to evaluate the future performance of the arch dams, the space–time monitoring model for the centroid is established by introducing its initial coordinates based on the traditional statistical models. Afterwards, a novel centroid prediction model is developed based on the least squares support vector machine to balance the empirical risk and generalization ability of the statistical regression models. The input factors of the prediction model are determined in advance by principal component analysis to eliminate the multi-collinearity and reduce the computational complexity. The model construction and validation of the centroid deformation method are implemented on the world’s highest arch dam through evaluating its structural behavior and predicting the development trend. The results can provide strong technical support to better grasp its performance during long-term operation.