Scholarlink

<p>The evaluation of uniaxial compressive strength (UCS) for rock is a crucial factor in guaranteeing the safety of subterranean mining endeavours. The presence of cracks in the rock typically reduces its compressive strength, and these cracks may act as potential weak points that are prone to damage. To determine the UCS of fractured weakly cemented rocks, a data-driven hybrid approach using machine learning was proposed in this study. First, a database of UCS for fractured weakly cemented rocks was created using particle flow code (PFC) numerical simulation software. This database includes 983 rock specimens based on the usual fracture geometric parameters, such as the fracture angle, width, length, number, and spacing. A novel nested cross-validation support vector machine-based interval prediction (LSSVM-NCV-ABKDE) model was developed to obtain the UCS of weakly cemented rocks. The performance of the proposed model was further compared using three traditional decision regression trees (DRT), multilayer perceptron (MLP), and support vector regression (SVR). The results show that the LSSVM-NCV-ABKDE model exhibits higher accuracy with a root mean square error (RMSE) of 0.336, mean absolute error (MAE) of 0.192, and coefficient of determination (R2) of 0.957. Notably, the fracture length and fracture angle are identified as two most influential parameters for UCS prediction based on the proposed experienced UCS prediction equation. This study can serve as a reliable reference for rapid assessment of the UCS of fractured weakly cemented rocks.</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

Predicted uniaxial compressive strength of fractured weakly cemented rock by LSSVM-NCV-ABKDE-based data-driven method

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月20日

查看全文
  • Yi-Dan Sun,
  • Chao Li,
  • Qiu-Yang Bi,
  • Yu Yang,
  • Guo-Bin Yang,
  • Chu Chen,
  • Hong-Zhi Cui

摘要

The evaluation of uniaxial compressive strength (UCS) for rock is a crucial factor in guaranteeing the safety of subterranean mining endeavours. The presence of cracks in the rock typically reduces its compressive strength, and these cracks may act as potential weak points that are prone to damage. To determine the UCS of fractured weakly cemented rocks, a data-driven hybrid approach using machine learning was proposed in this study. First, a database of UCS for fractured weakly cemented rocks was created using particle flow code (PFC) numerical simulation software. This database includes 983 rock specimens based on the usual fracture geometric parameters, such as the fracture angle, width, length, number, and spacing. A novel nested cross-validation support vector machine-based interval prediction (LSSVM-NCV-ABKDE) model was developed to obtain the UCS of weakly cemented rocks. The performance of the proposed model was further compared using three traditional decision regression trees (DRT), multilayer perceptron (MLP), and support vector regression (SVR). The results show that the LSSVM-NCV-ABKDE model exhibits higher accuracy with a root mean square error (RMSE) of 0.336, mean absolute error (MAE) of 0.192, and coefficient of determination (R2) of 0.957. Notably, the fracture length and fracture angle are identified as two most influential parameters for UCS prediction based on the proposed experienced UCS prediction equation. This study can serve as a reliable reference for rapid assessment of the UCS of fractured weakly cemented rocks.