Risk probabilistic assessment of ultrahigh arch dams through regression panel modeling on deformation behavior
Erfeng Zhao, Chengqing Wu
Abstract
The deformations of ultrahigh arch dams can comprehensively indicate the dynamic variations of their structural behavior to judge the normal or not for timely discovering anomalies. First, the panel data features on the deformation behavior are extracted to effectively indicate the overall structural evolution of the Jinping I arch dam combing with the time series and the cross-section series. Afterwards, a regression panel model (RPM) on the multi-dimensional variables is proposed to model the deformation panel data consisted of multi-monitoring points synchronously. Subsequently, an innovative functional relationship between the measured values and the real-time risk probabilistic function is established due to the RPM estimation accuracy. In order to estimate the risk probability of the whole arch dam, the Copula function is used to build a multivariate joint probability distribution function to indicate the correlation among the random variables. The proposed methods are validated by an application on the Jinping I arch dam to evaluate its risk probability, which explores a novel approach for the arch dam safety assessment.