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)
黄金矿山岩体质量分级知识库与PLS简化预测模型
- 李书强|刘志祥|刘伟军
- 作者单位：
- 赤峰柴胡栏子黄金矿业有限公司|中南大学资源与安全工程学院|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（52374107）
详细信息：
- 作者简介：
- 李书强(1990—)，男，工程师，从事采矿技术研究工作；E-mail：1552094009@qq.com
- 通讯作者：
- isnull
- PDF下载
Knowledge base for gold deposit rock mass quality grading and PLS simplified prediction model
- English Author：
- Chifeng Chaihulanzi Gold Mining Co.,Ltd.|School of Resources and Safety Engineering,Central South University|School of Resources and Safety Engineering,Central South University
- Unit：
- 摘要
- 在线预览
- 参考文献
摘要:
针对黄金矿山工程岩体特征，分析了岩石单轴抗压强度、RQD值、节理结构面状态、节理结构面间距、地下水状态、节理结构面方向?怨こ逃跋旌偷赜αχ嫡7个主要因素对岩体稳定性的影响，对7个指标进行修正，建立了地下矿山M-RMR岩体质量评价指标体系。采用M-RMR岩体质量评价指标体系划分了焦家金矿直属矿区、寺庄矿区和望儿山矿区工程岩体质量等级，建立了焦家金矿地下矿山岩体质量与其影响因素的神经网络知识库模型，达到了焦家金矿工程岩体质量智能分级的目的。为简化M-RMR指标体系中指标数量，更利于实际应用，采用变量投影重要性指标VIP对7个指标所携带信息量的大小进行排序，并逐个删除不重要的指标，利用单因变量的偏最小二乘回归方法（PLS1）建立了精简指标的简化预测模型。简化预测模型可使用较少的评价指标对岩体质量给出准确的分级，具有实际使用价值。
关键词:
黄金矿山;岩体质量分级；岩体稳定性；神经网络；知识库模型；简化模型；偏最小二乘回归方法
Abstract:
This study addresses the characteristics of rock masses in gold mine engineering,analyzing 7 key factors affecting rock mass stability:rock uniaxial compressive strength,RQD value,joint structural face conditions,joint structural face spacing,groundwater conditions,effect of joint structural face orientation on engineering,and on-situ stress value.These 7 indicators were adjusted to establish the rock mass quality M-RMR safety evaluation system.Using the M-RMR system,the engineering rock mass quality grades were classified for the Jiaojia Gold Mine-s directly managed mining area,Sizhuang mining area,and Wang-ershan mining area.Furthermore,a neural network knowledge base model was developed to correlate the underground rock mass quality at Jiaojia Gold Mine with its influencing factors,achieving intelligent grading of the rock mass quality for engineering purposes.To simplify the M-RMR indicator system for easier practical application,variable importance projection (VIP) was used to rank the information carried by the 7 indicators,allowing the removal of unimportant variables one by one.The simplified prediction model was built using partial least squares regression of single dependent variables (PLS1).This simplified model can accurately grade rock mass quality using fewer evaluation indicators,demonstrating practical application value.
Keywords:
gold mine;rock mass quality grading;rock mass stability;neural network;knowledge base model;simplified model;partial least squares regression
