Scholarlink

<p>Precise grade evaluation is an important step in the mining planning process. Geometrical approaches and geostatistical techniques are well established methods for grade estimation. Machine learning models are an emerging area and need more research in this domain. In this paper, random forests (RFs), XGBoost, and gradient boosting machine (GBM) are used to estimate ore grades in an Indian copper deposit. The models have been trained with 3132 number of samples containing their coordinates (east, north, and altitude) and ore grade (Copper). The data was split into 80% (3132 samples) for training, 10% (391 samples) for testing and 10% (391 samples) for validation. Performance was assessed using Mean Squared Error (MSE) and proportion of the variation (R<sup>2</sup>). The RFs has achieved MSE 0.0372 and R<sup>2</sup> 0.72. But XGboost and GBM models produce MSE 0.0385, R<sup>2</sup> 0.68 and MSE 0.0485, R<sup>2</sup> 0.54 respectively. In this study, we found that RFs model can predict the copper ore grades effectively for this type of deposits.</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

Grade Estimation Using Machine Learning Techniques in Indian Copper Deposits

  • 期刊论文

  • 发表日期：

    2025年2月14日

查看全文
  • Tapan Dey,
  • Gopinath Samanta

摘要

Precise grade evaluation is an important step in the mining planning process. Geometrical approaches and geostatistical techniques are well established methods for grade estimation. Machine learning models are an emerging area and need more research in this domain. In this paper, random forests (RFs), XGBoost, and gradient boosting machine (GBM) are used to estimate ore grades in an Indian copper deposit. The models have been trained with 3132 number of samples containing their coordinates (east, north, and altitude) and ore grade (Copper). The data was split into 80% (3132 samples) for training, 10% (391 samples) for testing and 10% (391 samples) for validation. Performance was assessed using Mean Squared Error (MSE) and proportion of the variation (R2). The RFs has achieved MSE 0.0372 and R2 0.72. But XGboost and GBM models produce MSE 0.0385, R2 0.68 and MSE 0.0485, R2 0.54 respectively. In this study, we found that RFs model can predict the copper ore grades effectively for this type of deposits.