Scholarlink

<p>This study introduces a novel hybrid framework that integrates principal component analysis (PCA) with advanced machine learning (ML) models for accurate prediction of rock fragmentation in surface mining. The framework aims to enhance operational efficiency, reduce environmental impact, and improve mine safety by optimizing blast design parameters. Four ML models—Random Forest (RF), Artificial Neural Network (ANN), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost)—were combined with PCA to reduce data dimensionality, eliminate redundancy, and improve model generalization. A dataset comprising 125 blasting cases was used, with 70% for training and 30% for testing. The performance of standalone and PCA-enhanced models was evaluated using multiple statistical metrics, including R<sup>2</sup>, RMSE, MAE, Adjusted R<sup>2</sup>, and Performance Index (PI). Among the models, PCA-RF demonstrated the highest predictive accuracy (R<sup>2</sup> = 0.995, RMSE = 0.011), outperforming other configurations. The findings reveal that PCA significantly enhances model robustness and reduces computational load. Moreover, the proposed PCA-ML models contribute to safer and more sustainable mining operations by enabling real-time fragmentation forecasting, reducing explosive consumption, and minimizing fly-rock, ground vibration, and air overpressure. This hybrid approach presents a scalable, cost-effective solution for data-driven blasting optimization in modern mining practices.</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

PCA-integrated machine learning framework for predicting rock fragmentation in blasting operations

  • 期刊论文

  • 发表日期：

    2025年8月9日

查看全文
  • Lassana P. Dukuly,
  • Megha Gupta,
  • Sufyan Ghani,
  • Waseem Akram

摘要

This study introduces a novel hybrid framework that integrates principal component analysis (PCA) with advanced machine learning (ML) models for accurate prediction of rock fragmentation in surface mining. The framework aims to enhance operational efficiency, reduce environmental impact, and improve mine safety by optimizing blast design parameters. Four ML models—Random Forest (RF), Artificial Neural Network (ANN), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost)—were combined with PCA to reduce data dimensionality, eliminate redundancy, and improve model generalization. A dataset comprising 125 blasting cases was used, with 70% for training and 30% for testing. The performance of standalone and PCA-enhanced models was evaluated using multiple statistical metrics, including R2, RMSE, MAE, Adjusted R2, and Performance Index (PI). Among the models, PCA-RF demonstrated the highest predictive accuracy (R2 = 0.995, RMSE = 0.011), outperforming other configurations. The findings reveal that PCA significantly enhances model robustness and reduces computational load. Moreover, the proposed PCA-ML models contribute to safer and more sustainable mining operations by enabling real-time fragmentation forecasting, reducing explosive consumption, and minimizing fly-rock, ground vibration, and air overpressure. This hybrid approach presents a scalable, cost-effective solution for data-driven blasting optimization in modern mining practices.