Scholarlink

<p>Microbially induced calcite precipitation (MICP) has acquired recognition as a promising approach for enhancing the geotechnical properties of sands. Predicting the unconfined compressive strength (UCS) of microbially induced calcite precipitation-treated sands is challenging due to the intricate interplay of influencing factors. This study introduces a hybrid machine learning model integrating gorilla troops optimizer—extreme gradient boosting, extreme gradient boosting, random forest, multi-expression programming, and decision tree for predicting the UCS of MICP-treated sands. The developed models are trained and validated using a comprehensive dataset comprising various input parameters, including MICP treatment conditions, sand properties, and calcite content. The results demonstrate the superiority of the gorilla troops optimizer—extreme gradient boosting over all individual models, with notable enhancements in predictive accuracy and robustness, having an <i>R</i><sup>2</sup> of 0.986. Shapley additive explanations results show that calcite content, initial void ratio, and uniformity coefficient are the primary predictors of UCS. Partial dependence plot results reveal nonlinear relationships between these factors and UCS.</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

Prediction of Unconfined Compressive Strength of Sands Treated with Microbially-Induced Calcite Precipitation Using a Hybrid Machine Learning Model

  • 期刊论文

  • 发表日期：

    2025年8月29日

查看全文
  • Husna Usman,
  • Majid Khan,
  • Waseem Akhtar Khan

摘要

Microbially induced calcite precipitation (MICP) has acquired recognition as a promising approach for enhancing the geotechnical properties of sands. Predicting the unconfined compressive strength (UCS) of microbially induced calcite precipitation-treated sands is challenging due to the intricate interplay of influencing factors. This study introduces a hybrid machine learning model integrating gorilla troops optimizer—extreme gradient boosting, extreme gradient boosting, random forest, multi-expression programming, and decision tree for predicting the UCS of MICP-treated sands. The developed models are trained and validated using a comprehensive dataset comprising various input parameters, including MICP treatment conditions, sand properties, and calcite content. The results demonstrate the superiority of the gorilla troops optimizer—extreme gradient boosting over all individual models, with notable enhancements in predictive accuracy and robustness, having an R2 of 0.986. Shapley additive explanations results show that calcite content, initial void ratio, and uniformity coefficient are the primary predictors of UCS. Partial dependence plot results reveal nonlinear relationships between these factors and UCS.