A true triaxial strength criterion for rocks by gene expression programming

Rock strength is a crucial factor to consider when designing and constructing underground projects. This study utilizes a gene expression programming (GEP) algorithm-based model to predict the true triaxial strength of rocks, taking into account the influence of rock genesis on their mechanical behavior during the model building process. A true triaxial strength criterion based on the GEP model for igneous, metamorphic and magmatic rocks was obtained by training the model using collected data. Compared to the modified Weibols-Cook criterion, the modified Mohr-Coulomb criterion, and the modified Lade criterion, the strength criterion based on the GEP model exhibits superior prediction accuracy performance. The strength criterion based on the GEP model has better performance in R-2, RMSE and MAPE for the data set used in this study. Furthermore, the strength criterion based on the GEP model shows greater stability in predicting the true triaxial strength of rocks across different types. Compared to the existing strength criterion based on the genetic programming (GP) model, the proposed criterion based on GEP model achieves more accurate predictions of the variation of true triaxial strength (sigma(1)) with intermediate principal stress (sigma(2)). Finally, based on the Sobol sensitivity analysis technique, the effects of the parameters of the three obtained strength criteria on the true triaxial strength of the rock are analysed. In general, the proposed strength criterion exhibits superior performance in terms of both accuracy and stability of prediction results. (c) 2023 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V.

Automated summary

This study uses a gene expression programming (GEP) algorithm-based model to predict the true triaxial strength of rocks and obtains a strength criterion for different types of rocks. The strength criterion based on the GEP model exhibits superior prediction accuracy and stability compared to existing criteria.