| 基于随机森林算法的大尹格庄金矿床三维成矿预测 |
| |
| 引用本文： | 陈进,毛先成,刘占坤,邓浩.基于随机森林算法的大尹格庄金矿床三维成矿预测[J].大地构造与成矿学,2020(2):231-241. |
| |
| 作者姓名： | 陈进 毛先成 刘占坤 邓浩 |
| |
| 作者单位： | 中南大学地球科学与信息物理学院;中南大学有色金属成矿预测与地质环境监测教育部重点实验室 |
| |
| 基金项目： | 国家重点研发计划项目(2017YFC0601503);国家自然科学基金项目(41772349、41472301)联合资助。 |
| |
| 摘 要： | 大数据及机器学习技术在解决各行各业的复杂非线性关系问题方面已经体现出巨大的优势。本文尝试将随机森林(RF)算法引入三维成矿预测领域来开展研究,以胶东大尹格庄金矿为研究对象,在构建招平断裂(地质体)三维模型的基础上,通过各种空间分析方法提取控制矿体形成的若干控矿地质因素特征值,进而获取成矿空间中控矿地质因素分布值,最后将矿区钻孔立体单元化形成采样数据集并利用RF算法对矿区开展三维矿体定位预测,结果表明:决策树棵数M=800、属性个数K=7是最优参数,能获得总体精度97.32%和kappa系数0.6292的综合分类精度;RF算法的分类精度要优于支持向量机(SVM)算法和多层感知器(MP)算法。RF算法对大尹格庄金矿开展的三维矿体定位预测取得了较好效果,并在矿区深边部预测了7个三维找矿靶区,证明大数据技术在矿产资源定位预测方面具有巨大的应用前景。
|
| 关 键 词： | 大数据技术 机器学习 随机森林算法 控矿因素 三维成矿预测 大尹格庄金矿床 |
Three-dimensional Metallogenic Prediction Based on Random Forest Classification Algorithm for the Dayingezhuang Gold Deposit |
| |
| CHEN Jin,MAO Xiancheng,LIU Zhankun,DENG Hao.Three-dimensional Metallogenic Prediction Based on Random Forest Classification Algorithm for the Dayingezhuang Gold Deposit[J].Geotectonica et Metallogenia,2020(2):231-241. |
| |
| Authors: | CHEN Jin MAO Xiancheng LIU Zhankun DENG Hao |
| |
| Institution: | (School of Geosciences and Info-Physics,Central South University,Changsha 410083,Hunan,China;Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring,Ministry of Education,Central South University,Changsha 410083,Hunan,China) |
| |
| Abstract: | Big data and machine learning technologies have shown great advantages in solving complex nonlinear relationships in various industries.The core of mineral resource location prediction can be attributed to the nonlinear relationship between mineralization distribution and ore-controlling geological factors.However,it is difficult to meet the objective,reliable and quantitative relationship by relying on traditional logic hypothesis or statistical analysis.Therefore,this paper attempts to introduce random forest(RF)algorithm into the field of three-dimensional metallogenic prediction.Specifically,the Dayingezhuang gold deposit in Jiaodong is studied,based on the construction of the 3D model of the Zhaoping fault(geological body).Firstly,the characteristic values of some ore-controlling geological factors that control the formation of orebodies were extracted by various spatial analysis methods.Secondly,the distribution values of ore-controlling geological factors in the ore-forming space were obtained,and finally 3D orebody location prediction in the mining area was carried out using the RF algorithm on the sampling dataset from the stere-unitization of the drills.The results showed that the number of decision trees M=800 and the number of attributes K=7 are the optimal parameters.The overall classification accuracy can achieve an overall accuracy of 97.32%and a kappa coefficient of 0.6292;the classification accuracy of RF is better than that of the support vector machine(SVM)and multi-layer perceptron(MP).The RF algorithm yielded a good result in the 3D orebody location prediction of the Dayingezhuang gold deposit,predicting seven 3D prospecting targets in the deep-edge space of the mining area.Our study showed that big data technologies have great application prospects in mineral resources evaluation and prediction. |
| |
| Keywords: | big data technology machine learning random forest algorithm ore-controlling geological factors threedimensional metallogenic prediction Dayingezhuang gold deposit |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
