Scholarlink

<p>Heat stress conditions in underground mines significantly impact miners’ health, safety, and productivity. Therefore, precise heat stress prediction is pivotal for implementing appropriate mitigation strategies. This study presents five distinct regression models—Multilinear Regression (MLR), Support Vector Regression (SVR), Ridge Regression (RR), Decision Tree Regression (DTR), and Random Forest Regression (RFR)—to estimate the wet bulb globe temperature (WBGT) in underground metalliferous mines. A total of 310 datasets were collected from field studies at two underground mines. The regression models comprise three independent variables: air velocity, dry bulb temperature, and wet bulb temperature. The dependent variable is WBGT. A total of 250 data sets were used for training and testing of the models, while 60 were used for validation. The prediction performance of the models is evaluated by two error criteria, namely the correlation coefficient (R²) and root mean square error (RMSE). The R² values for MLR, SVR, RR, DTR, and RFR were 0.92, 0.93, 0.92, 0.91, and 0.93, respectively. The SVR and RFR models outperformed the other models with the lowest RMSE of 0.63. The findings of this study indicate that SVR and RFR accurately estimate heat stress conditions in underground metalliferous mines.</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

Evaluation of AI-Based Regression Models for Predicting Heat Stress (WBGT) in Underground Mines

  • 期刊论文

  • 发表日期：

    2025年11月13日

查看全文
  • Durga Nookaraju Gadhi,
  • Mahesh Kumar Shriwas,
  • Ramakrishna Morla,
  • B. Paluchamy

摘要

Heat stress conditions in underground mines significantly impact miners’ health, safety, and productivity. Therefore, precise heat stress prediction is pivotal for implementing appropriate mitigation strategies. This study presents five distinct regression models—Multilinear Regression (MLR), Support Vector Regression (SVR), Ridge Regression (RR), Decision Tree Regression (DTR), and Random Forest Regression (RFR)—to estimate the wet bulb globe temperature (WBGT) in underground metalliferous mines. A total of 310 datasets were collected from field studies at two underground mines. The regression models comprise three independent variables: air velocity, dry bulb temperature, and wet bulb temperature. The dependent variable is WBGT. A total of 250 data sets were used for training and testing of the models, while 60 were used for validation. The prediction performance of the models is evaluated by two error criteria, namely the correlation coefficient (R²) and root mean square error (RMSE). The R² values for MLR, SVR, RR, DTR, and RFR were 0.92, 0.93, 0.92, 0.91, and 0.93, respectively. The SVR and RFR models outperformed the other models with the lowest RMSE of 0.63. The findings of this study indicate that SVR and RFR accurately estimate heat stress conditions in underground metalliferous mines.