KMS XINJIANG INSTITUTE OF ECOLOGY AND GEOGRAPHY，CAS
| 基于证据权和模糊逻辑的三维成矿预测研究—以东天山红海矿床为例 | |
| Alternative Title | Research on three-dimensional metallogenic prediction based on weights of evidence and fuzzy logic: A case study of the Honghai deposit, Eastern Tianshan |
| 陶金涛 | |
| Subtype | 硕士 |
| Thesis Advisor | 袁峰 ; 张楠楠 |
| 2020-06-30 | |
| Degree Grantor | 中国科学院大学 |
| Place of Conferral | 北京 |
| Degree Discipline | 工程硕士 |
| Keyword | 证据权 模糊逻辑 三维地质建模 成矿预测 ROC 曲线 Weights of evidence Fuzzy logic 3D geological modeling Prospectivity modeling Receiver operating characteristics (ROC) |
| Abstract | 随着易于寻找的地表矿、浅部矿的不断减少，隐伏矿、深部矿、难识别矿已经成为我国大部分地区找矿的主要对象。面对当前深边部找矿的需求，将传统的二维成矿预测方法扩展到三维空间进行三维成矿预测的研究成为近年来矿产资源预测领域的一大亮点。证据权方法和模糊逻辑方法作为基于数据驱动和基于知识驱动的成矿预测方法中使用最为广泛的方法之一，研究这两种方法在三维成矿预测中的应用对于当前深部成矿预测工作有着重要的意义。在三维成矿预测应用中， 对于证据权方法和模糊逻辑方法具体参数的研究较少，并且对于预测模型的有效评估的研究还比较薄弱。 本文以新疆东天山卡拉塔格地区的红海 VMS（ Volcanogenic Massive Sulfide） 矿床为实例，展开基于证据权和模糊逻辑的三维成矿预测研究，并利用 ROC（ Receiver OperatingCharacteristics） 曲线及相关指标对预测模型进行了评估比较。 本文的主要工作和成果如下：（ 1）收集红海矿床的原始地质数据资料包括平面地形地质图、中断平面图、勘探线剖面图以及钻孔编录数据等，并利用 GIS 软件对数据进行了编辑整合。根据研究区的矿床地质特征，通过 GOCAD（ Geological Object Computer AidedDesign） 软件完成了红海矿区黄铁绢英岩化、凝灰岩、熔岩、安山岩、英安岩、断层以及破碎带的三维实体模型的构建，并在此基础上采用三维缓冲区分析实现了各控矿因素距离场的定量计算。（ 2） 结合红海矿床的地质特征和提取的三维控矿因素， 分别利用三维证据权方法和三维模糊逻辑方法，获得了红海矿区的后验概率分布图和模糊概率分布图。根据预测结果的捕获效率曲线进行了找矿有利单元的圈定， 结果表明证据权方法在识别红海矿床深部隐伏矿体方面比模糊逻辑方法具有更好的表现，在红海已知矿体的南部有着较大的找矿潜力，并且利用额外的钻孔数据对这一结果进行了验证。（ 3） 通过对 ROC 曲线中验证数据集正负样本数目对评价结果的影响的研究，表明 ROC 曲线能有效地评估验证三维成矿预测方法。 利用 ROC 曲线和约登指数等性能评价指标对证据权预测模型和模糊逻辑预测模型进行了评估比较，评估比较的结果表明两种模型都是有效的，但证据权预测模型性能总体上优于模糊逻辑预测模型。基于三维成矿预测的实际需求，构建系统架构，设计功能模块，利用 Python 开发了三维成矿预测系统 V1.0。 |
| Other Abstract | With the number of outcropping and near-surface mineral deposits that areeasily discovered decreasing, concealed, deep-seated and indiscernible mineraldeposits have become the main exploration targeting of most areas in China. In theface of current demand for deep-side prospecting, the study of extending theconventional 2D methods to 3D metallogenic prediction has become a highlight inthe field of mineral resource prediction in recent years. Weights of evidence methodand fuzzy logic method, as one of the most widely used methods in data-driven andknowledge-based metallogenic prediction, respectively, are of great significance forthe current deep metallogenic prediction.In the application of 3D prospectivity modelling, there are rare researches onthe specific parameters of the weights of evidence and fuzzy logic, and theresearches on the effective evaluation of prospectivity models are still weak. In thisstudy, 3D prospectivity modeling based on weights of evidence and fuzzy logic,respectively, was applied to the Honghai VMS(Volcanogenic Massive Sulfide)deposit in eastern Tianshan, northwestern China, and ROC(Receiver OperatingCharacteristics) curves and relevant indexes were used to evaluate and compare theperformance of the two methods. The main work and achievements of this studywere presented as follows:(1) The original geological data of the Honghai deposit were collected,including geological maps, geological plans, cross sections and boreholes, and wereedited and integrated by GIS software. According to the geological characteristics ofthe Honghai deposit, 3D solid geological models of phyllic alteration, tuff, lava,andesite, dacite, fault and fracture zone were constructed using theGOCAD(Geological Object Computer Aided Design) software. On this basis, thedistance field of each ore-controlling factor was calculated quantitatively by usingthree-dimensional buffer analysis.(2) Combined with the geological characteristics of the Honghai deposit and the extracted ore-controlling factors, a posterior probability distribution map and afuzzy probability distribution map were generated by using 3D weights of evidencemethod and 3D fuzzy logic method, respectively. Capture efficiency curves wereused to delineate favorable units for ore prospecting in the prospectivity maps. Thefavorable units indicate that weights of evidence method has better performancethan fuzzy logic method regarding identification of the known mineralization in thestudy area. It has a great potential in the areas south of the known orebody, and extraboreholes were used to verify the result.(3) The study of the effect of the numbers of positive and negative samples onthe ROC curve indicates that ROC curve can effectively evaluate and verify the 3Dmetallogenic prediction method. The two prospectivity models were evaluated andcompared by using ROC curve, Youden index and other evaluation indexes. Theevaluation results indicate the effectiveness of both methods for the 3D prospectivitymodeling of the Honghai deposit. However, weights of evidence generallyperformed better than fuzzy logic for identification of the concealed and deep-seatedHonghai deposit. Based on the actual demand for 3D metallogenic prediction, asystem architecture was constructed, function modules were designed, and 3Dmetallogenic prediction system V1.0 finally was developed by python. |
| Subject Area | 地球探测与信息技术 |
| Language | 中文 |
| Document Type | 学位论文 |
| Identifier | http://ir.xjlas.org/handle/365004/15449 |
| Collection | 中国科学院新疆生态与地理研究所 研究系统 |
| Affiliation | 中国科学院新疆生态与地理研究所 |
| First Author Affilication | 中国科学院新疆生态与地理研究所 |
| Recommended Citation GB/T 7714 | 陶金涛. 基于证据权和模糊逻辑的三维成矿预测研究—以东天山红海矿床为例[D]. 北京. 中国科学院大学,2020. |
| Files in This Item: | There are no files associated with this item. | |||||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment