基于集成学习法预测岩石单轴抗压强度

打开文本图片集

中图分类号：TP18；TU45 文献标志码：A

Abstract：This study employs ensemble learning methods to determine rock strength.A comprehensive database encompassing various rock properties -including porosity，wave velocity，and elastic modulus -was established.Ensemble learning techniques were thenutilized to efectively capture the nonlinear relationships between these factors and rock strength，enabling accurate prediction. Following parameter selection from the database，a predictive model dataset was constructed for training and hyperparameter optimization.The trained ensemble models were subsequently applied to predict rock uniaxial compresive strength. Results demonstrated that the stacking method achieved the highest predictionaccuracy among all ensemble models，while the integration of the automated hyperparameter framework Optuna significantly enhanced model performance.This approach provides an eficient and reliable method for rock strength prediction.

Key words：uniaxial compressive strength of rock ； rock wave velocity；ensemble learning；stacking method

岩石的抗压强度是岩土、地质等工程领域的重要参数[1-2]，广泛用于隧道、坝体及岩坡的稳定性分析。（剩余11280字）