中国科技核心期刊
美国化学文摘社（CAS）数据库
美国EBSCO学术数据库
日本科学技术振兴机构数据库(JST)
基于Optuna-RF智能模型的采矿方法优选
- 作者：
许杨丰1，陈煜新2，周健2，邱贤阳2，田志刚1
- 作者单位：
- （1. 中金岭南有色金属股份有限公司凡口铅锌矿；2. 中南大学资源与安全工程学院）
基金项目：
- 国家自然科学基金项目（52374152）
详细信息：
- 作者简介：
- 许杨丰（1986—），男，工程师，从事矿山技术管理工作；E⁃mail：349246040@qq. com
- PDF下载
Optimal selection of mining methods based on the Optuna-RF intelligent model
- English Author：
Xu Yangfeng¹, Chen Yuxin², Zhou Jian², Qiu Xianyang², Tian Zhigang¹
- Unit：
- （1. Fankou Lead-Zinc Mine, Zhongjin Lingnan Non‑ferrous Metals Co., Ltd.; 2. School of Resources and Safety Engineering, Central South University）
- 摘要
- 在线预览
- 参考文献
摘要:
随着全球工业需求的增加，高效且可靠的采矿方法选择变得至关重要。研究通过结合先进的机器学习技术和自动化超参数优化框架，探索提升采矿方法选择的科学性和精确性。通过收集和分析多种采矿条件及技术经济指标数据，构建了一个基于Optuna的优化随机森林模型。该模型的优化过程专注于调整RF模型的 4个关键超参数，旨在提高模型的预测准确率及其泛化能力。试验结果表明，优化后的RF模型在训练集和测试集的分类准确率均明显提高了，Optuna-RF模型优于未优化的RF模型。这一结果验证了自动超参数优化在提升机器学习模型泛化能力和预测精度方面的关键作用，为采矿方法的智能选择提供了一种有效的技术路径。
关键词:
人工智能；采矿方法；分类预测；智能模型；自动化；超参数；随机森林；Optuna
Abstract:
With the growing global demand for industrial resources, the selection of efficient and reliable miningmethods has become increasingly critical. This study explores ways to enhance the scientific and accurate selection ofmining methods by integrating advanced machine learning techniques with an automated hyperparameter optimizationframework. Based on the collection and analysis of various mining conditions and techno⁃economic indicators, anoptimized random forest model based on Optuna was developed. The optimization process focused on tuning 4 keyhyperparameters of the RF model to improve its predictive accuracy and generalization performance. Experimentalresults showed that the optimized RF model evidently improved classification accuracy on both the training and testingdatasets. The Optuna-RF model significantly outperformed the unoptimized model. These results demonstrate the crucialrole of automated hyperparameter optimization in enhancing the generalization and predictive capabilities of machinelearning models and provide an effective technical approach for the intelligent selection of mining methods.
Keywords:
artificial intelligence; mining method; classification prediction; intelligent model; automation; hyper⁃parameters; random forest; Optuna
