Improved approach for vibration-based structural health monitoring of arch dams during seismic events and normal operation

Vibration-based structural health monitoring for arch dams is influenced by various environmental effects, such as water depth and air temperature. These effects do not cause structural damage; therefore, their effects need to be minimized or removed from models. Moreover, arch dams are often located in earthquake-prone areas; thus, an improved approach is proposed herein to adapt to the aforementioned situations. In this study, an arch dam in China was used for a case study. A long-term continuous structural health monitoring system was proposed to operate under seismic and environmental excitation. An autoregressive exogenous model was applied for modal identification under seismic excitation. An automatic procedure based on stochastic subspace identification and density-based spatial clustering of applications with noise was carried out for modal identification under environmental excitation. This study focused on the identification of the first three-order modes. The seismic events were then analyzed. The natural frequencies before, during, and after seismic events were the same within a certain margin of error. This indicates that earthquakes did not cause major seismic damage to the arch dam. In addition, natural frequencies were continuously identified under environmental excitation. The relationship between natural frequency and water depth forms a quadratic curve, and the modal damping ratio increases with increasing water depth. Finally, this paper presents a random forest model to minimize environmental effects on natural frequencies. After minimizing the influence of non-structural factors, the natural frequency exhibited a weak increasing trend over time.

Automated summary

This study focused on vibration-based structural health monitoring for arch dams, considering environmental effects and seismic excitation. A case study was conducted on a Chinese arch dam, and modal identification methods were proposed for seismic and environmental excitation. The study analyzed the seismic events and the relationship between natural frequency and water depth, and presented a random forest model to minimize environmental effects on natural frequencies.