Abstract: In current nuclear engineering, the use of commercial finite element software for seismic mechanical computation and analysis of a large number of equipment items suffers from several limitations, including long computation time, heavy consumption of computational resources, and the inability to reuse historical data. To address these issues, this paper innovatively applies six ensemble learning models, including CatBoost and LightGBM, to rapidly predict key mechanical response parameters—such as modal frequencies, stresses, and foundation loads of typical horizontal and vertical pressure vessels in nuclear engineering under multiple load types(e.g., self-weight, internal pressure, seismic loads, nozzle loads) and multi-condition combinations. A large-scale, high-dimensional dataset was automatically constructed via Latin Hypercube Sampling and ANSYS Batch modules, comprising 8,451 samples with 144 effective features for horizontal vessels and 11,903 samples with 172 effective features for vertical vessels. Model performance was comprehensively evaluated and analyzed using two metrics: the coefficient of determination (R²) and mean absolute percentage error (MAPE). The results show that the CatBoost model achieves high accuracy and robustness in predicting most physical quantities, which is of great significance for meeting the demand for rapid-response engineering design.