Scholarlink

Point Load Strength Index (PLI) is considered an indirect method of estimation of Unconfined Compressive Strength (UCS) as given in IS 8764. It states a constant conversion factor of 22 from PLI to UCS irrespective of rock type, origin, its zonal presence and other properties. Errors up to 100% can be expected if an arbitrary ratio is chosen resulting in under/over estimation of rock strength in terms of UCS. In this study, rock from a particular site (Kamalanga, Dist. Dhenkanal, State: Odisha, India) has been collected for experimental analysis. The rock samples have been tested suitably as per relevant IS codes to determine the conversion factor. The physical testing has revealed that the conversion factor is quite different from the value mentioned in IS 8764. In view of the discrepancy noted in the value of conversion factor, the present study aims to develop a machine learning model while exploring Linear Regression, ANN Regression as well as Polynomial Regression strategies so that this conversion factor can be reliably predicted using some crucial physical parameters of the rock like Bulk Density, Water Content, Porosity, Specific Gravity, Saturated UCS, Point Load Strength Index, and RQD. According to the experimentation, ANN-based Regression model is found to be most suitable for reliable prediction of the conversion factor as it scores a promising 91% R-Squared value.

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

A Machine Learning Based Model for Predicting the Ratio of UCS and Point Load Strength of a Typical Rock in the State of Odisha, India

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sudip Nath,
  • Sibapriya Mukherjee,
  • Sourav Saha

摘要

Point Load Strength Index (PLI) is considered an indirect method of estimation of Unconfined Compressive Strength (UCS) as given in IS 8764. It states a constant conversion factor of 22 from PLI to UCS irrespective of rock type, origin, its zonal presence and other properties. Errors up to 100% can be expected if an arbitrary ratio is chosen resulting in under/over estimation of rock strength in terms of UCS. In this study, rock from a particular site (Kamalanga, Dist. Dhenkanal, State: Odisha, India) has been collected for experimental analysis. The rock samples have been tested suitably as per relevant IS codes to determine the conversion factor. The physical testing has revealed that the conversion factor is quite different from the value mentioned in IS 8764. In view of the discrepancy noted in the value of conversion factor, the present study aims to develop a machine learning model while exploring Linear Regression, ANN Regression as well as Polynomial Regression strategies so that this conversion factor can be reliably predicted using some crucial physical parameters of the rock like Bulk Density, Water Content, Porosity, Specific Gravity, Saturated UCS, Point Load Strength Index, and RQD. According to the experimentation, ANN-based Regression model is found to be most suitable for reliable prediction of the conversion factor as it scores a promising 91% R-Squared value.