Scholarlink

<p>Flyrock, a dangerous by-product of blasting operations in surface mining, presents substantial threats to human safety, infrastructure, and the surrounding environment. This paper introduces a hybrid machine learning framework to enhance the accuracy of flyrock prediction and support safer, more sustainable mining practices. Advanced gradient boosting techniques, namely light gradient boosting machine and CatBoost (category gradient boosting), were combined with four powerful metaheuristic optimization algorithms, namely Archimedes optimization algorithm, jellyfish search optimizer (JSO), Harris hawks optimizer, and geometric mean optimizer, resulting in eight hybrid predictive models. These models were trained and validated using a 252-record dataset from the Sungun Copper Mine and were assessed through multiple performance metrics, including error indices, Taylor diagrams, relative absolute error-cumulative frequency plots, and violin plots. Among the developed models, JSO-CatBoost achieved the highest predictive performance, demonstrating minimal error and reduced uncertainty. SHapley Additive exPlanations and empirical cumulative distribution function analysis revealed that burden was the most influential factor in flyrock occurrence. This paper underscores the potential of optimized ensemble learning models to improve blasting design, minimize environmental hazards, and enhance operational safety in surface mining. This work contributes to the broader field of earth system modeling and environmental hazard mitigation through the integration of data-driven and interpretative artificial intelligence methods.</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

Intelligent Prediction of Flyrock Hazards in Surface Mining Using Optimized Gradient Boosting Models

  • 期刊论文

  • 发表日期：

    2025年8月16日

查看全文
  • Mohammad Matin Rouhani,
  • Mahdi Hasanipanah,
  • Xin Yin,
  • Iman Ahmadianfar,
  • Hesam Dehghani

摘要

Flyrock, a dangerous by-product of blasting operations in surface mining, presents substantial threats to human safety, infrastructure, and the surrounding environment. This paper introduces a hybrid machine learning framework to enhance the accuracy of flyrock prediction and support safer, more sustainable mining practices. Advanced gradient boosting techniques, namely light gradient boosting machine and CatBoost (category gradient boosting), were combined with four powerful metaheuristic optimization algorithms, namely Archimedes optimization algorithm, jellyfish search optimizer (JSO), Harris hawks optimizer, and geometric mean optimizer, resulting in eight hybrid predictive models. These models were trained and validated using a 252-record dataset from the Sungun Copper Mine and were assessed through multiple performance metrics, including error indices, Taylor diagrams, relative absolute error-cumulative frequency plots, and violin plots. Among the developed models, JSO-CatBoost achieved the highest predictive performance, demonstrating minimal error and reduced uncertainty. SHapley Additive exPlanations and empirical cumulative distribution function analysis revealed that burden was the most influential factor in flyrock occurrence. This paper underscores the potential of optimized ensemble learning models to improve blasting design, minimize environmental hazards, and enhance operational safety in surface mining. This work contributes to the broader field of earth system modeling and environmental hazard mitigation through the integration of data-driven and interpretative artificial intelligence methods.