Abstract: Reservoir dams are major lifeline projects, and how to quickly and effectively conduct seismic damage assessment of reservoir dams after an earthquake is of great significance for the development of rescue programs and post-disaster restoration. In order to quickly and accurately assess the damage status of reservoir dams that have been subjected to earthquakes., we select the details of the earthquake damage of each reservoir dam in Wenchuan 8.0 magnitude earthquake, and construct a sample dataset by combining the structural characteristics of the dams and the intensity of the earthquake. The missing values of the samples are processed using k-nearest neighbor interpolation, and the relevance of the sample features is judged. In this paper, we propose a rapid assessment model of earthquake damage of reservoir dams based on the gradient boosting tree algorithm.Four hyperparametric optimization methods, namely Grid Search (GS), Particle swarm optimization (PSO), Bayesian optimization (BO) and HyperBand Search (HS), are used for the Gradient Boosting Tree (GBDT) regression algorithm for parameter optimization, and the models were compared according to their performance metrics (coefficient of determination R2, root mean square error RMSE, mean error MAE) and ranked in terms of feature importance. The results show that the BO-GBDT model can assessment the degree of earthquake damage of reservoir dams with the shortest time consumption as well as higher accuracy, its regression coefficient R2 is as high as 0.98, and the characteristic importance score indicates that maximum crack width is the most influential factor. Comparison of the assessment results using this model with an earth dam earthquake damage assessment model based on an improved empirical statistical model showed further improvement in accuracy, which verified the reliability of the model in the application of rapid investigation and assessment of post-earthquake damage to reservoir dams.