Scholarlink

<p>Drilling and blasting techniques are cost effective and efficient methods for the drivage of development headings in underground mines and tunnelling projects. Pull and overbreak are major challenges associated with these techniques, and it is important to maximise pull and minimise overbreak during blasting operations. Accurate prediction of these factors before blasting can significantly improve operational efficiency. In this study, multiple linear regression and machine learning models were employed to predict pull and overbreak. The K-nearest neighbor (KNN), Random Forest (RF) and Gradient Boosting Regressor (GBR) models were used for this purpose. Five input parameters namely number of holes, average hole depth, total explosive fired in the round, maximum charge weight per delay and uniaxial compressive strength of the rock, were collected from the experimental site. A total dataset of 155 points was compiled and split into training and testing sets in a 70: 30 ratio. The performance of the developed models was evaluated using root mean square error (RMSE) and the coefficient of determination (R<sup>2</sup>). Among the models, GBR exhibited the highest accuracy in predicting pull (RMSE = 1.09 and R<sup>2</sup> = 0.94) and overbreak (RMSE = 4.11% and R<sup>2</sup> = 0.95), indicating superior predictive capability compared to KNN and RF. These results suggest that GBR based predictive models can be effectively used for estimating pull and overbreak in underground development face blasting.</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 overbreak and pull using machine learning approaches in the development face blasting of underground metalliferous mines

  • 期刊论文

  • 发表日期：

    2026年3月25日

查看全文
  • Ashish Kumar Vishwakarma,
  • Vivek Kumar Himanshu,
  • Kaushik Dey,
  • Saket Kumar

摘要

Drilling and blasting techniques are cost effective and efficient methods for the drivage of development headings in underground mines and tunnelling projects. Pull and overbreak are major challenges associated with these techniques, and it is important to maximise pull and minimise overbreak during blasting operations. Accurate prediction of these factors before blasting can significantly improve operational efficiency. In this study, multiple linear regression and machine learning models were employed to predict pull and overbreak. The K-nearest neighbor (KNN), Random Forest (RF) and Gradient Boosting Regressor (GBR) models were used for this purpose. Five input parameters namely number of holes, average hole depth, total explosive fired in the round, maximum charge weight per delay and uniaxial compressive strength of the rock, were collected from the experimental site. A total dataset of 155 points was compiled and split into training and testing sets in a 70: 30 ratio. The performance of the developed models was evaluated using root mean square error (RMSE) and the coefficient of determination (R2). Among the models, GBR exhibited the highest accuracy in predicting pull (RMSE = 1.09 and R2 = 0.94) and overbreak (RMSE = 4.11% and R2 = 0.95), indicating superior predictive capability compared to KNN and RF. These results suggest that GBR based predictive models can be effectively used for estimating pull and overbreak in underground development face blasting.