基于XGBoost模型岩体可爆性研究  

Study on explosibility of rock mass based on XGBoost model

作　　者：吴凌峰 周宗红[2] 孙伟 Wu Lingfeng;Zhou Zonghong;Sun Wei(Jinping Chang an Mining Co.,Ltd.;School of Land and Resources Engineering,Kunming University of Science and Technology)

机构地区：[1]金平长安矿业有限公司 [2]昆明理工大学国土资源工程学院

出　　处：《黄金》2024年第2期21-23,32,共4页Gold

基　　金：国家自然科学基金项目(52264019,51864023)。

摘　　要：岩体可爆性是衡量岩体爆破难易程度的一个重要指标,准确对岩体可爆性评价能够为合理爆破设计提供依据。选取岩石密度、单轴抗压强度、岩石抗拉强度、岩石脆性指数、动载强度和完整性系数等作为岩体可爆性数据集的指标,采用Z-Score方法标准化岩体可爆性数据集,消除量纲对模型预测影响,分别采用朴素贝叶斯、支持向量机和XGBoost模型进行岩体可爆性分级,结果表明:采用XGBoost模型能够准确评价岩体的可爆性,为岩体可爆性评价提供一种新的方法。Rock mass explosibility is an important index to measure the difficulty of rock mass blasting,and an accurate evaluation of rock mass explosibility can provide a basis for reasonable blasting design.In this paper,rock density,uniaxial compressive strength,rock tensile strength,rock brittleness index,dynamic load strength,and integrity coefficient are selected as the indicators of rock mass explosibility data set.The data set of rock mass explosibility is standardized by Z-Score,and the influence of dimension on model prediction is eliminated.Naive Bayes,support vector machine,and XGBoost models are used to classify rock mass explosibility.The results show that XGBoost model can accurately evaluate rock mass explosibility and provide a new method for rock mass explosibility evaluation.

关 键 词：爆破 岩体可爆性 可爆性分级 XGBoost 机器学习算法 

分 类 号：TD235[矿业工程—矿井建设]