Leveraging Artificial Neural Networks to Predict Igneous Rock Strength Parameters from Petrological Contents

Accurately determining rock strength characteristics, such as uniaxial compressive strength (UCS), Brazilian tensile strength (BTS), and Los Angeles Abrasion (LAA), through conventional methods is both time-consuming and resource-intensive. To address this challenge, this study develops efficient artificial neural network (ANN) models optimized with Levenberg-Marquardt (LM), Bayesian Regularization (BR), and Scaled Conjugate Gradient (SCG) algorithms to predict UCS, BTS, and LAA from petrographic properties, advancing civil engineering applications. A total of 100 dolerite samples were analyzed to assess their strength and petrographic characteristics, with the ANNs trained on the three aforementioned algorithms. The results indicate that the BR model achieved the highest accuracy, with a correlation coefficient (R) of 0.9999 and a root mean square error (RMSE) of 0.3164. The LM model also demonstrated strong performance with an R-value of 0.9997 and an RMSE of 0.8619. The BR and LM models significantly outperformed the SCG model, which had an R-value of 0.9954 and an RMSE of 2.398. Sensitivity analysis identified plagioclase and chlorite as the most influential factors in predicting rock mechanical parameters. The effectiveness of the BR and LM techniques highlights their potential to offer substantial time and cost savings in rock strength prediction.