黄建,黄家龙,赵坦.矿石体重不同分析方法的对比探讨——以长山锌金多金属矿床为例[J].地质找矿论丛,2021,36(3):378-387
矿石体重不同分析方法的对比探讨——以长山锌金多金属矿床为例
Parartive study on analysis methods for volumetric weight of ore -a case of Changshan Zn-Au poly metallic deposit
投稿时间:2021-02-03  修订日期:2021-06-07
DOI:10.6053/j.issn.1001-1412.2021.03.015
中文关键词:  长山锌金多金属矿  矿石体重  多元线性回归模型  神经网络模型
英文关键词:Changshan Zn-Au poly metallic deposit  volumetric weight of ore  multiple regression model  neural network model
基金项目:
作者单位
黄建 华东冶金地质勘查研究院, 合肥 230088 
黄家龙 华东冶金地质勘查研究院, 合肥 230088 
赵坦 华东冶金地质勘查研究院, 合肥 230088 
摘要点击次数: 30
全文下载次数: 59
中文摘要:
      矿石体重的准确获取,对降低矿床储量计算误差十分重要。文章以长山锌金多金属矿为例,建立矿石体重与Cu、Pb、Zn、S等元素之间的数学模型。首先探讨矿石体重和元素之间的数学关系,并证明当矿石中有多种硫化物建立回归模型时,不需要对全硫进行扣减;在此基础上,分别建立2种多元线性回归模型和1种神经网络模型,并对3种模型拟合精度进行对比。结果表明,以矿石体重倒数为因变量要优于以矿石体重为因变量建立的回归模型,而神经网络模型明显优于回归模型。
英文摘要:
      Accurate acquisition of volumetric weight is very important to reduce calculation errors of ore reserves. Mathematic model of volumetric weight and elements of Cu, Pb, Zn and S in ore of Changshan Zn-Au poly metallic deposit in the east China is established to explore the math relation between the volumetric weight and the elements. It is proved that if Multi-sulfides occur in ore establishment of the ore regression model dose not need deduction of total sulfur. Based on the math model are established 2 model of multiple linear regression and 1 neural network model. Fitting accuracy of the three models is compared. The regression model whose variable is reciprocal of volumetric weigh is better than that whose variable is volumetric weight and the neural network model much better than the regression model
查看全文  查看/发表评论  下载PDF阅读器
关闭