.png){kind=icon}
.png){kind=icon}
Return
An advanced data reliability assessment method with application to high rockfill dam deformation analysis
.png){kind=icon}
PRE
AI
J
C
Y
P
DOI：10.1016/j.ress.2025.112130.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Deformation is the most direct indicator of the structural and safety status of high rockfill dams. Accurately evaluating the reliability of deformation monitoring data and understanding deformation patterns and their controlling factors are critical for the construction of dam prediction model and safety assessment. However, existing data reliability methods often neglect the influence of physical mechanisms of deformation, which may result in inaccurate identification of abnormal data caused by local structural changes and sensor failures. Therefore, this study proposes a data-driven and physical collaborative data reliability assessment method in high rockfill dams. This method integrates Time-delay Embedding Principal Component Analysis (TdEPCA) for feature extraction, Mahalanobis distance (Md) for correlation feature construction, and Dempster-Shafer (DS) evidence theory for abnormal probability fusion, forming the TdEPCA-Md-DS model to comprehensive decision-making on data reliability. The PB high rockfill dam is used as the case study, where the model accurately identifies the abnormal sensor locations and the data reliability is determined as 84.295%, validating its effectiveness. Meanwhile, the sensitivity analysis of the principal component contribution ensures its applicability and good robustness in multiple scenarios. The proposed method provides a practical and robust tool for enhancing deformation monitoring and supporting safety management of high rockfill dams. On the basis of identifying and removing abnormal data, the deformation characteristics of the dam during construction, initial impoundment, and operational stages are analyzed in detail, revealing the main factors controlling deformation in rockfill dams.
Journal
R
IF:
11
Papers:
9.0K
Citations:
4.2W
Organization
Cited Papers