Seismic Hazard Loss Assessment of Reservoir Dams Based on Random Forest Algorithm

Reservoir dams can have a significant impact on social production and daily life in the aftermath of earthquakes. Therefore, it is crucial to implement prompt and effective seismic damage assessment and emergency disposal programs for reservoir dams following seismic events. This paper combines the information of reservoir damage data of the Wenchuan earthquake, takes the reservoir dam as the main research body, analyzes the type of damage, formulates the water conservancy project seismic damage evaluation level, and selects 17 influence factors to establish a sample data set. The sample data are preprocessed first, stratified sampling is used for sample splitting cross validation, and parameter tuning is performed through random search. The optimized random forest algorithm model was compared with three regression algorithms to plot the R2 (determination coefficient) curve. The optimized random forest had the highest determination coefficient (0.84), indicating that it performed the best in predicting the reservoir dam seismic damage level. The loss ratio of the main project of the reservoir dam was calculated using the predicted earthquake damage level from the model. Subsequently, the direct economic loss was determined by considering the total cost of the main project within the designated assessment area

Algorithm · Forensic engineering · Hazard · Hazard analysis · Machine learning · Random forest · Reliability engineering · Seismic hazard · Seismology · Anomaly Detection Techniques and Applications · Computer Science · Dam Engineering and Safety · Engineering · Environmental Science · Geology

• Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance

  CJ Willmott, C J Willmott et al.•Climate Research•2005

  Incluir item

• High-accuracy estimation method of typhoon storm surge disaster loss under small sample conditions by information diffusion model coupled with machine learning models

  Open Access•Xuexue Du, Xiaomin Li et al.•International Journal of Disaster…•2022

  Incluir item