期刊文献+

基于注意力卷积神经网络的焦家金矿带三维成矿预测及构造控矿因素定量分析 被引量:7

Three-dimensional prospectivity mapping and quantitative analysis of structural ore-controlling factors in Jiaojia Au ore-belt with attention convolutional neural networks
下载PDF
导出
摘要 焦家金矿带是我国重要的金矿产地,随着开采深度不断增加,深部找矿已成为目前工作重点,为此,从数据科学视角出发,利用深度学习技术,开展焦家矿带三维成矿预测及控矿因素定量分析工作。在建立三维地质模型和控矿指标基础上,构建引入CBAM注意力机制模块的卷积神经网络模型,从初始控矿指标中抽取具有矿化指示性的信息特征,建立焦家断裂面与矿化定位间的非线性关联关系,并与其他方法构建的成矿预测模型进行对比,验证本文方法构建的模型的准确性与可靠性。利用DeepLIFT方法解构深度网络特征,明确各控矿指标输入对网络输出的贡献,以此定量分析焦家矿带控矿因素对金成矿的影响。研究结果表明:焦家断裂距离场对成矿影响最显著,其次为坡度和坡度变化率,形态起伏度对成矿影响较弱;在矿带深部2000~3000 m圈定找矿有利靶区3处,其中,纱岭勘查区矿体深部延伸部位和曲家勘查区北段深部具有较大找矿潜力,焦家与曲家勘查区连接部位深部可能存在新的矿化富集区。 Jiaojia gold ore-belt is an important gold producing area in China.With the increase of mining depth,deep prospecting has become the focus of current work.Starting from the perspective of data science,deep learning was used to carry out 3D metallogenic prediction and quantitative analysis of ore-controlling factors in Jiaojia ore-belt.Based on the 3D model and conceptional models,a convolution neural network model was constructed with CBAM attention mechanism module,and then information features were extracted from initial factors to establish a reliable association between Jiaojia fault and gold distribution,and the prospecting results verified the accuracy and reliability comparied with other prospectivity methods.To clarify the contribution of each ore-controlling factor to the network output and thus the influence of ore-controlling factors on gold mineralization,DeepLIFT method was used to inspect the deep network.The results show that the distance to Jiaojia fault has the most significant influence on mineralization,followed by the slope and dip-transition factors,and the shape fluctuation has a relatively weak influence on mineralization.The 3D prediction model highlights three favorable prospecting targets at the depth of 2000-3000 m.The dip direction along Shaling exploration area and the north part of Qujia exploration area have great prospecting potential,and another target in the deep joint part between the Jiaojia and Qujia exploration areas,it is likely that there exists a new gold enrichment zone.
作者 邓浩 魏运凤 陈进 刘占坤 喻姝研 毛先成 DENG Hao;WEI Yunfeng;CHEN Jin;LIU Zhankun;YU Shuyan;MAO Xiancheng(Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Central South University,Changsha 410083,China;School of Geosciences and Info-physics,Central South University,Changsha 410083,China)
出处 《中南大学学报(自然科学版)》 EI CAS CSCD 北大核心 2021年第9期3003-3014,共12页 Journal of Central South University:Science and Technology
基金 国家重点研发计划项目(2017YFC0601503) 国家自然科学基金资助项目(41972309,41772349,42030809,42072325) 中南大学中央高校基本科研业务费专项资金资助项目(2020zzts672)。
关键词 三维成矿预测 焦家金矿带 注意力卷积神经网络 构造控矿因素分析 3D metallogenic prediction Jiaojia Au ore-belt attention convolutional neural network analysis of structure-controlled factors
  • 相关文献

参考文献19

二级参考文献360

共引文献829

同被引文献130

引证文献7

二级引证文献11

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部