Safety Monitoring of High Arch Dams in Initial Operation Period Using Vector Error Correction Model

• Published in: Rock mechanics and rock engineering Vol. 51; no. 8; pp. 2469 - 2481
• Main Authors: Liang, Guohe, Hu, Yu, Li, Qingbin
• Format: Journal Article
• Language: English
• Published: Vienna Springer Vienna 01.08.2018; Springer Nature B.V
• Subjects: Arch dams; Autocorrelation; Body temperature; Civil Engineering; Cointegration analysis; Dam safety; Dams; Data processing; Deformation; Dimensions; Earth and Environmental Science; Earth Sciences; Economic models; Error correction; Error correction & detection; Geophysics/Geodesy; Mathematical models; Model accuracy; Monitoring; Original Paper; Principal components analysis; Safety; Statistical analysis; Statistical models; Temperature; Time series; Safety monitoring of dams; Cointegration analysis; Hydrostatic-seasonal-time model; Vector error correction model
• ISSN: 0723-2632; 1434-453X
• DOI: 10.1007/s00603-017-1287-y

Conventional statistical models for dam safety monitoring often require long-term, continuous and stationary monitoring time series, which are difficult to fulfill in the initial operation period of the dams. In this study, special attention was given to the nonstationarity and lack of monitoring time series, and a vector error correction model was proposed for the safety monitoring of arch dams in their initial operation period. Principal component analysis was used in the data preprocessing stage to extract uncorrelated representative temperature trends of the dam body from hundreds of multisensor temperature records to reduce variable dimensions. Then, the vector error correction model was proposed, to take into account the cointegration between the structural responses and the environmental variables. The model was further extended in order to take into account the autocorrelation and cross-correlation among multiple structural responses. The established model performed better in terms of fitting and prediction accuracy compared with existing models and provided better forecast even when limited observations were available. The proposed method was successfully implemented to analyze the deformation of the Xiluodu arch dam in southwest China.