A hybrid data-mechanism-driven cloud model for spatiotemporal deformation-informed safety assessment of high arch dams
Yingrui Wu, Fei Kang, Jing Chen, Ming Yu, Kaiyuan Zheng
Abstract
The safety of high arch dams has emerged as a critical concern in hydraulic engineering, owing to their widespread application. Traditional dam safety assessment methods often fail to adequately consider the spatiotemporal distribution characteristics of dam deformations and their dynamic variations, thereby hindering a comprehensive and accurate reflection of the dam’s actual operational status. To address these limitations and comprehensively evaluate the overall safety of high arch dams, this study introduces a data-mechanism hybrid driven safety assessment cloud model. The dynamic time warping (DTW) approach, combined with hierarchical clustering techniques, is employed to perform spatiotemporal clustering of dam deformation sequences, thus enabling precise identification of deformation patterns within each partition. Subsequently, a FEM-HTT hybrid prediction model is developed by integrating finite element method (FEM) with the Hydrostatic-Thermal-Time (HTT) monitoring theory, which facilitates accurate and reliable deformation predictions. Finally, the dam’s overall operational status is evaluated using multi-layer data fusion and cloud model theory. The proposed method is applied to the safety assessment of a super high arch dam, and the results demonstrate that the method effectively evaluates the dam’s operational state, providing a theoretical foundation for its comprehensive safety evaluation.