当前位置: X-MOL 学术Nat. Resour. Res. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Machine-Learning-Aided Determination of Post-blast Ore Boundary for Controlling Ore Loss and Dilution
Natural Resources Research ( IF 5.4 ) Pub Date : 2021-07-05 , DOI: 10.1007/s11053-021-09914-5
Zhi Yu 1 , Xiuzhi Shi 1 , Jian Zhou 1 , Yonggang Gou 1 , Dijun Rao 1 , Xiaofeng Huo 1
Affiliation  

In the process of open-pit bench blasting for many mines, rock fragments move in the direction of loose space after fragmentation under explosive energy, leading to ore distribution differences before and after blasting. Considering that there is no simple and inexpensive method for post-blast ore boundary determination, a machine-learning-aided determination method and a corresponding evaluation system are proposed. In this regard, 95 datasets with nine variables were investigated using different kinds of predictive models: support vector regression, the Gaussian process (GP), extreme learning machine, and two metaheuristic algorithms. By evaluating the predictive performance using three performance metrics, absolute error comparison, and Taylor diagram, a hybrid model composed of whale optimization algorithm (WOA) and GP, namely WOA–GP, was found to be the best precision model. Then, a blast block with 55 blast holes was assumed and was analyzed using the WOA–GP model and the established evaluation system. It was found that using the post-blast ore boundary to guide the shovel can decrease the ore loss rate by 33.1% and the ore dilution rate by 57.9%, meaning that the methodology has great significance for improving resource recovery and increasing economic benefits. In addition, the mixing of high-grade and low-grade rock fragments during the geological modeling process may also cause ore loss and dilution. This article provides a new methodology for post-blast ore boundary determination, which can inspire development of other ore loss and dilution management techniques.



中文翻译:

机器学习辅助确定爆炸后矿石边界以控制矿石损失和稀释

许多矿山在露天平台爆破过程中,岩屑在爆炸能量作用下破碎后向松散空间方向移动,导致爆破前后矿石分布存在差异。考虑到爆后矿石边界确定没有简单、廉价的方法,提出了一种机器学习辅助确定方法和相应的评价系统。在这方面,使用不同类型的预测模型对 95 个具有 9 个变量的数据集进行了研究:支持向量回归、高斯过程 (GP)、极限学习机和两种元启发式算法。通过使用三个性能指标、绝对误差比较和泰勒图来评估预测性能,由鲸鱼优化算法(WOA)和 GP 组成的混合模型,即 WOA-GP,被发现是最好的精度模型。然后,假设一个具有 55 个爆破孔的爆破块,并使用 WOA-GP 模型和建立的评估系统进行分析。结果表明,采用爆破后矿石边界引导铲车,可使矿石损失率降低33.1%,矿石稀释率降低57.9%,对提高资源回收率和提高经济效益具有重要意义。此外,地质建模过程中高品位和低品位岩石碎片的混合也可能造成矿石的流失和稀释。本文为爆破后矿石边界确定提供了一种新方法,可以启发其他矿石损失和稀释管理技术的发展。假设有一个具有 55 个爆破孔的爆破块,并使用 WOA-GP 模型和已建立的评估系统进行分析。结果表明,采用爆破后矿石边界引导铲车,可使矿石损失率降低33.1%,矿石稀释率降低57.9%,对提高资源回收率和提高经济效益具有重要意义。此外,地质建模过程中高品位和低品位岩石碎片的混合也可能造成矿石的流失和稀释。本文为爆破后矿石边界确定提供了一种新方法,可以启发其他矿石损失和稀释管理技术的发展。假设有一个具有 55 个爆破孔的爆破块,并使用 WOA-GP 模型和已建立的评估系统进行分析。结果表明,采用爆破后矿石边界引导铲车,可使矿石损失率降低33.1%,矿石稀释率降低57.9%,对提高资源回收率和提高经济效益具有重要意义。此外,地质建模过程中高品位和低品位岩石碎片的混合也可能造成矿石的流失和稀释。本文为爆破后矿石边界确定提供了一种新方法,可以启发其他矿石损失和稀释管理技术的发展。结果表明,采用爆破后矿石边界引导铲车,可使矿石损失率降低33.1%,矿石稀释率降低57.9%,对提高资源回收率和提高经济效益具有重要意义。此外,地质建模过程中高品位和低品位岩石碎片的混合也可能造成矿石的流失和稀释。本文为爆破后矿石边界确定提供了一种新方法,可以启发其他矿石损失和稀释管理技术的发展。结果表明,采用爆破后矿石边界引导铲车,可使矿石损失率降低33.1%,矿石稀释率降低57.9%,对提高资源回收率和提高经济效益具有重要意义。此外,地质建模过程中高品位和低品位岩石碎片的混合也可能造成矿石的流失和稀释。本文为爆破后矿石边界确定提供了一种新方法,可以启发其他矿石损失和稀释管理技术的发展。地质建模过程中高品位和低品位岩石碎片的混合也可能导致矿石损失和稀释。本文为爆破后矿石边界确定提供了一种新方法,可以启发其他矿石损失和稀释管理技术的发展。地质建模过程中高品位和低品位岩石碎片的混合也可能导致矿石损失和稀释。本文为爆破后矿石边界确定提供了一种新方法,可以启发其他矿石损失和稀释管理技术的发展。

更新日期:2021-07-05
down
wechat
bug