On-line diagnosis method of crack behavior abnormality in concrete dams based on fluctuation of sequential parameter estimates

Abstract

Research on on-line diagnosis method of crack behavior abnormality in concrete dams can provide support for timely grasping abnormality state of the crack itself and achieving real-time monitoring of the dam safety. Considering that samples of crack effects in concrete dams increase actually over monitoring time, a superiority criterion for the on-line diagnosis is determined so as to detect the abnormality moments timely and reliably. By integrating the safety monitoring statistical model of crack effect variable with change point theory, a fluctuation method of regression coefficients is established for the on-line diagnosis. In addition, each abnormality moment is detected by the cumulative sum of regression model residuals. Results indicate that abnormality of crack behavior in concrete dams can be characterized by structural instability of crack monitoring model. And causes of crack behavior abnormality can be analyzed by the established method, which will play an important role in dam safety monitoring. Further, taking the crack in a concrete gravity-arch dam as an example, the scientific rationality and validity of the established on-line diagnosis method are confirmed.