Scholarlink

<p>One of the significant hazards in the underground mining industry is roof fall triggered due to variations in the geomechanical condition. It inflicts enormous losses in terms of work time, injuries and fatalities. The most adopted approach to countering such hazards relies on visual inspection and expert knowledge. Machine Learning technologies have the potential to improve the safety, operational efficiency and productivity of the mining industry. The current work invigorates an ML-based hazard management system with a smaller dataset. It presents an approach that measures and assesses the performance of different machine learning classifiers to classify the risk associated with roof falls accurately. The classifiers include Bi-layer Neural Networks (BNN), Bagged Trees (BT), Random Subspace (RS) classifiers using Linear Discriminant Learning and K-Nearest Neighbour (KNN). According to the findings, the Bagged trees, the subspace KNN, and the Bilayer NN all reached the highest possible classification accuracy of 90%. Bagged trees and subspace KNN classifiers, on the other hand, have the potential to be superior in terms of successfully transferring the learning outcome. The work highlights the significance of machine learning in improving geological hazard management as a valuable knowledge addition to human expertise. The proposed approach is poised to play a crucial role in autonomous mining operations, particularly in scenarios where hazard identification relies on human expertise.</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

Leveraging ensemble learning and neural networks enhancing underground mine roof safety

  • 期刊论文

  • 发表日期：

    2025年6月16日

查看全文
  • Jitendra Pramanik,
  • Arun Kumar Sahoo,
  • Abhaya Kumar Samal,
  • Singam Jayanthu

摘要

One of the significant hazards in the underground mining industry is roof fall triggered due to variations in the geomechanical condition. It inflicts enormous losses in terms of work time, injuries and fatalities. The most adopted approach to countering such hazards relies on visual inspection and expert knowledge. Machine Learning technologies have the potential to improve the safety, operational efficiency and productivity of the mining industry. The current work invigorates an ML-based hazard management system with a smaller dataset. It presents an approach that measures and assesses the performance of different machine learning classifiers to classify the risk associated with roof falls accurately. The classifiers include Bi-layer Neural Networks (BNN), Bagged Trees (BT), Random Subspace (RS) classifiers using Linear Discriminant Learning and K-Nearest Neighbour (KNN). According to the findings, the Bagged trees, the subspace KNN, and the Bilayer NN all reached the highest possible classification accuracy of 90%. Bagged trees and subspace KNN classifiers, on the other hand, have the potential to be superior in terms of successfully transferring the learning outcome. The work highlights the significance of machine learning in improving geological hazard management as a valuable knowledge addition to human expertise. The proposed approach is poised to play a crucial role in autonomous mining operations, particularly in scenarios where hazard identification relies on human expertise.