Scholarlink

<p>The strength enhancement in rocks under the influence of high strain rates can be utilized to its full potential in various rock mechanics processes like drilling, blasting, mining, and tunneling. However, intrinsic properties of rocks, such as porosity, affect the strength enhancement in rocks. This paper aims to understand the combined effect of strain rate and porosity on the mechanical behavior of rocks using a Machine Learning (ML) based approach. The main input parameters include sample porosity, strain rate, sample height, diameter, and dry density. Among various ML models, XGBoost achieved the best performance with R² values of 0.990 (compression) and 0.978 (indirect tension) for strength prediction, and 0.790 (compression) and 0.877 (indirect tension) for predicting dynamic increase factors (DIFs), along with the lowest error values. To further investigate the influence of various input parameters, Permutation Feature Importance (PFI), SHapley Additive exPlanation (SHAP), and Partial Dependence Plot (PDP) analyses were conducted. The results indicate that porosity and strain rate are the overall primary factors. SHAP and PDP analyses reveal a positive correlation between strength and strain rate, while a negative correlation is observed with porosity. Notably, porosity is the most dominant parameter for predicting compressive strength, while strain rate is more significant in predicting indirect tensile strength. However, for CDIF and TDIF, strain rate is the dominant factor, followed by porosity.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Rate-dependent compressive and tensile strength of rocks: from prediction to explanation using machine learning

  • 期刊论文

  • 发表日期：

    2026年6月30日

查看全文
  • Peerzadi Arzeena Imtiyaz,
  • Shruti Upadhyaya,
  • Shwetabh Yadav

摘要

The strength enhancement in rocks under the influence of high strain rates can be utilized to its full potential in various rock mechanics processes like drilling, blasting, mining, and tunneling. However, intrinsic properties of rocks, such as porosity, affect the strength enhancement in rocks. This paper aims to understand the combined effect of strain rate and porosity on the mechanical behavior of rocks using a Machine Learning (ML) based approach. The main input parameters include sample porosity, strain rate, sample height, diameter, and dry density. Among various ML models, XGBoost achieved the best performance with R² values of 0.990 (compression) and 0.978 (indirect tension) for strength prediction, and 0.790 (compression) and 0.877 (indirect tension) for predicting dynamic increase factors (DIFs), along with the lowest error values. To further investigate the influence of various input parameters, Permutation Feature Importance (PFI), SHapley Additive exPlanation (SHAP), and Partial Dependence Plot (PDP) analyses were conducted. The results indicate that porosity and strain rate are the overall primary factors. SHAP and PDP analyses reveal a positive correlation between strength and strain rate, while a negative correlation is observed with porosity. Notably, porosity is the most dominant parameter for predicting compressive strength, while strain rate is more significant in predicting indirect tensile strength. However, for CDIF and TDIF, strain rate is the dominant factor, followed by porosity.