Scholarlink

<p>Opencast mining accounts for over 95% of India’s coal production, with surface miners enabling continuous, selective mining; however, productivity drops abruptly in hard rocks conditions (60&#xa0;MPa). To date, cuttability research has largely focused on single conical picks, with limited work on harder-rock cutting using scaled drums with multi‑pick arrays. This study introduces an indigenous Rotary Cutting Rig (RCR) for block‑level rotary cutting experiments and develops a data‑driven framework to evaluate and optimize cutting drum performance. An experimental dataset was generated by varying cutting and rock parameters, and machine‑learning models were trained to predict peak and mean forces and specific energy. Random Forest, hyper-tuned with Bayesian optimization, provided reliable predictions, while SHAP and LIME explainability methods quantified the influence and non‑linear effects of cutting parameters and rock properties on performance parameters. The trained surrogate models were coupled with an enhanced Non‑dominated Sorting Genetic Algorithm‑III to maximize the mean‑to‑peak-force ratio while minimizing specific energy to achieve optimal cutting conditions. The optimal configuration was found at 35&#xa0;mm pick line spacing, 1.25&#xa0;m/min cutting speed, and 100&#xa0;rpm drum speed while cutting rock at a characterized P‑wave velocity of 917&#xa0;m/s and rebound hardness of 47 RI, obtained from grid‑wise block characterization. Confirmation tests demonstrated the repeatability of both the RCR experimentation and the predictive-optimization framework, indicating strong potential for operating regime selection and performance enhancement of cutting drums in medium-strength rock for surface miner applications.</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

Machine‑Learning‑Assisted Experimental Investigation and Multi‑objective Optimization of Cutting Drum Design and Operating Parameters for Surface Miners Using a Rotary Cutting Rig

  • 期刊论文

  • 发表日期：

    2026年7月24日

查看全文
  • Hemant Kumar,
  • L. A. Kumaraswamidhas,
  • V. M. S. R. Murthy

摘要

Opencast mining accounts for over 95% of India’s coal production, with surface miners enabling continuous, selective mining; however, productivity drops abruptly in hard rocks conditions (60 MPa). To date, cuttability research has largely focused on single conical picks, with limited work on harder-rock cutting using scaled drums with multi‑pick arrays. This study introduces an indigenous Rotary Cutting Rig (RCR) for block‑level rotary cutting experiments and develops a data‑driven framework to evaluate and optimize cutting drum performance. An experimental dataset was generated by varying cutting and rock parameters, and machine‑learning models were trained to predict peak and mean forces and specific energy. Random Forest, hyper-tuned with Bayesian optimization, provided reliable predictions, while SHAP and LIME explainability methods quantified the influence and non‑linear effects of cutting parameters and rock properties on performance parameters. The trained surrogate models were coupled with an enhanced Non‑dominated Sorting Genetic Algorithm‑III to maximize the mean‑to‑peak-force ratio while minimizing specific energy to achieve optimal cutting conditions. The optimal configuration was found at 35 mm pick line spacing, 1.25 m/min cutting speed, and 100 rpm drum speed while cutting rock at a characterized P‑wave velocity of 917 m/s and rebound hardness of 47 RI, obtained from grid‑wise block characterization. Confirmation tests demonstrated the repeatability of both the RCR experimentation and the predictive-optimization framework, indicating strong potential for operating regime selection and performance enhancement of cutting drums in medium-strength rock for surface miner applications.