Rock Deformation Parameter Estimation Method Based on Stacking Integrated Learning

Geomechanical parameters of rocks, such as Young's modulus (E), play an important role in oil drilling and other related engineering work as an important parameter for the analysis of stratigraphic stress characteristics and the evaluation of oil and gas reservoirs. However, laboratory and field measurements of Young's modulus are time-consuming and labor-intensive, and require specialized experimental equipment and personnel to complete. Therefore, this study aims to develop a new machine learning model that can efficiently and accurately predict the Young's modulus of rocks. The model utilizes Stacking integrated learning theory to fuse several single heterogeneous models such as XGBoost, RF, MLP, and KNN, and finally constructs a fusion model, Stacking-RF-MLP-KNN, which has a better model performance and higher prediction accuracy. In order to validate the prediction effect of the model, the model is compared with the individual single models. compared with each single model, and the acceptability of the above methods is tested by calculating statistical functions such as coefficient of determination (R2), mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE). The experimental results show that the Stacking-RF-MLP-KNN model developed in this study has the highest prediction accuracy (RMSE of 13.821, R2 of 0.813).