Journal Article

Predicting the elasticity modulus of sedimentary rocks using Deep Random Forest Optimization (DRFO) algorithm

Zhang, Maosheng, Teng, Fei, Azarafza, Mohammad, Li, Yican, Nanehkaran, Yaser A., Mao, Yimin

2024

Overview

The accurate determination of rock elasticity modulus is crucial for geomechanical analysis and reliable rock engineering designs. Traditional experimental methods have limitations in estimating elasticity modulus, prompting the adoption of artificial intelligence and data-driven techniques to develop adaptive and accurate predictive models. This study utilized the Deep Random Forest Optimization (DRFO) algorithm, a hybrid approach combining deep learning and random forest algorithms, to predict rock elasticity modulus. The dataset consisted of 350 sedimentary rock samples from various regions in Iran, including sandstone, limestone, marlstone, and mudstone. The performance of the predictive models was assessed using confusion matrices, statistical errors, and the coefficient of determination (R 2 ). The results revealed the superior performance of the DRFO model, exhibiting a remarkably low Mean Absolute Error (MAE) of 0.180 GPa, outperforming other models. The Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) values (0.026 and 0.161, respectively) confirmed the precision of DRFO’s predictions. DRFO demonstrated robustness and generalization capability, yielding excellent performance in both training and testing datasets. Moreover, accuracy and precision evaluation in the training dataset showed a high accuracy (0.97) and precision (0.97), indicating the reliability of DRFO in estimating rock elasticity modulus. The study underscores the significance of data-driven techniques, particularly the potential of DRFO in accurately predicting rock properties. It contributes valuable insights to the field of geotechnical engineering, aiding infrastructure design and ensuring the safety and stability of sedimentary rock-based structures. Further research can explore DRFO’s adaptability to different geological contexts and extend its application to other essential rock properties, advancing geotechnical and geological engineering practices. The integration of advanced data-driven approaches like DRFO can enhance rock mechanics understanding, facilitating sustainable engineering solutions for various geotechnical projects.

• Publisher: Springer Berlin Heidelberg,Springer Nature B.V
• Subject: Accuracy / Adaptability / Algorithms / Artificial intelligence / Biogeosciences / data collection / Datasets / Deep learning / Earth and Environmental Science / Earth Sciences / Elasticity / Engineering / Environmental engineering / Environmental Science and Engineering / Experimental methods / Geochemistry / Geology / Geomechanics / Geotechnical engineering / Hydrology/Water Resources / infrastructure / Iran / Limestone / Machine learning / Marlstone / mechanics / modulus of elasticity / Mudstone / Optimization / Original Article / Performance evaluation / Performance prediction / Prediction models / Rock mechanics / Rock properties / Root-mean-square errors / Sandstone / Sediment samples / Sedimentary rocks / Statistical analysis / Statistical models / Terrestrial Pollution / Training