Application of signal processing techniques in structural health monitoring of concrete gravity dams

Destruction of dams often has extreme financial consequences and sometimes results in fatalities. Therefore, structural health monitoring is a crucial issue. In this article, the renowned Pine Flat dam has been chosen for the finite element modeling. The main objective is damage detection based on behavior evaluation of intact and damaged dams when an earthquake is applied. Three well-known earthquake records including El Centro, Northridge, and Loma Prieta, respectively, with low, middle, and high frequencies are considered in this study. At the data generation stage, the damage is induced in the dam neck and heel through elasticity modulus reduction, and then any of the aforementioned earthquakes are applied to it. In addition, for more accommodation with practical situations, the observations are contaminated with random noise to cover an amount of uncertainty due to some realistic factors. Following that, the acceleration values of different nodes of both intact and damaged structures are saved in two data vectors. Using various methods, such as discrete-time Fourier transform (DTFT), wavelet, and Wiener transforms, the differences between intact and damaged data are investigated. To provide a quantitative and precise analysis of the efficiency of each method in terms of destruction detection, the standard deviation of variations is employed. The results demonstrate that the wavelet transform outperforms other algorithms. Moreover, it is shown that wavelet transform represents the best robustness regarding noise variance in recorded observations. © 2023, The Author(s), under exclusive licence to Springer Nature Switzerland AG.