| 基于SVM模型的胶东金矿成矿预测研究 |
| |
| 引用本文： | 张宏睿,张庆现.基于SVM模型的胶东金矿成矿预测研究[J].中国矿业,2020,29(S1). |
| |
| 作者姓名： | 张宏睿 张庆现 |
| |
| 作者单位： | 中国地质大学（北京）,山东省地质矿产勘查开发局 |
| |
| 基金项目： | 北京市优秀硕士学位论文指导老师科研项目 |
| |
| 摘 要： | 本文系统结合前人研究资料，进行控矿要素分析，总结金矿成矿规律，采用支持向量机模型（SVM）进行深部隐伏金矿预测。首先根据研究区矿床类型，得知该区金矿床为造山型，矿床严格受构造、岩浆岩等方面控制。建立了勘查模型，Au、Ag两种贵金属元素经多重分形滤波（S-A）分离得到的背景和异常，进而将各勘查标志转换为证据图层，采用高斯、线性、双曲正切SVM、逻辑回归模型、朴素贝叶斯模型进行成矿预测，最终生成成矿潜力图和累计预测面积比例与累计预测矿床数量图，共圈定远景区5个。结果表明高斯、线性、双曲正切核函数的SVM模型逻辑回归模型相差无几，且均优于证据权模型，表明支持向量机是一种具有广泛的应用前景的成矿预测方法。
|
| 关 键 词： | 成矿预测 支持向量机 金矿 核函数 数值模拟 |
| 收稿时间： | 2020/5/5 0:00:00 |
| 修稿时间： | 2020/6/18 0:00:00 |
Metallogenic prediction in jiaodong gold deposit based on SVM model |
| |
| ZHANG Hongrui and ZHANG Qingxian.Metallogenic prediction in jiaodong gold deposit based on SVM model[J].China Mining Magazine,2020,29(S1). |
| |
| Authors: | ZHANG Hongrui and ZHANG Qingxian |
| |
| Affiliation: | China University of Geosciences (Beijing),Shandong provincial bureau of geology and mineral resources |
| |
| Abstract: | Based on the previous research data, this paper systematically analyzes the ore-controlling factors, summarizes the metallogenic rules of gold deposits, and uses support vector machine (SVM) model to predict the hidden gold deposits in depth.Firstly, according to the types of ore deposits in the study area, it is known that the gold deposits in this area are orogenic, and the ore deposits are strictly controlled by structure and magmatic rocks.Exploration model is established, Au, Ag, two kinds of precious metal elements by the multifractal filter (S-A) isolated from background and anomaly, which converts the exploration marks to evidence layer and adopt gauss, linear and hyperbolic tangent SVM, the logistic regression model, simple bayesian model of metallogenic prediction, the resulting metallogenic potential graph and total area ratio and the accumulative prediction deposit amount figure, 5 were delineated vision area.The results show that the logistic regression models of gaussian, linear and hyperbolic tangent kernel functions are similar to each other, and all of them are better than the weight of evidence model. |
| |
| Keywords: | Mineral prospectivity mapping Support vector machine Gold mine Kernel function numerical simulation |
|
| 点击此处可从《中国矿业》浏览原始摘要信息 |
|
点击此处可从《中国矿业》下载全文 |
|
