KMS XINJIANG INSTITUTE OF ECOLOGY AND GEOGRAPHY，CAS
| 矿产资源预测判别系统关键技术的研发与应用 | |
| 杜茜诗慧 | |
| Subtype | 博士 |
| Thesis Advisor | 崔遥 ; 周可法 |
| 2018-06-05 | |
| Degree Grantor | 中国科学院大学 |
| Place of Conferral | 新疆乌鲁木齐 |
| Degree Discipline | 理学博士 |
| Keyword | 定量预测方法 判别函数 矿产资源 金矿 哈图-包古图地区 Quantitative prediction methods Discrimination functions Mineral Resources Gold deposits Hatu-Baogutu District |
| Abstract | 在信息化时代背景下，面对人均资源占有量与资源储量日益突出的矛盾，必须创新和突破新的成矿理论和关键技术，积极探索新方法新技术在矿产资源预测研究中的应用。高新技术的迅速发展，极大丰富了矿产资源预测理论和方法。每个预测方法和评价方法在实践应用中都有各自的优点和不足，但目前国际上还没有统一的规则用于选择最符合实际需要的方法。本论文从预测方法与评价方法的数学原理出发，从计算机科学的角度将数学模型与编程技术相结合，探索了矿产资源定量预测技术与预测结果判别技术，将 GIS 技术、数据库技术以及可视化技术耦合应用，最终构建了预测过程自动化与预测结果最优化的矿产资源预测判别系统，为矿产资源预测和评价提供了新技术。本文取得的主要进展包括：（1）矿产资源定量预测方法的选择引入了文献计量学的基本理论及其前沿技术方法，采用自然语言处理和文本挖掘技术对矿产资源预测的科学文本进行处理，共提取出 24 种矿产资源定量预测方法，并通过共现分析实现了对预测方法科学结构的可视化表达；采用双标图分析法确定了预测方法的热点与前沿，最终选择了证据权法、模糊逻辑、神经网络和支持向量机四种定量预测方法。（2）矿产资源定量预测关键技术的研发矿产资源定量预测技术是在详细研究了证据权法、模糊逻辑、BP 神经网络和支持向量机的基础上，将编程技术与数学模型充分结合，并解决了其中涉及的关键问题，包括条件独立性检验、模糊推理网络、神经网络参数设置以及交叉验证，完成了预测过程从非自动化向自动化转变的初步探索；矿产资源预测结果判别技术是在充分研究了预测结果评价方法的基础上，提出了基于评价方法的联合判别函数的概念，能够实现对预测结果快速、准确和客观地判别，并以 ROC 曲线和 P-A 图为例，分析了联合判别函数的具体表达形式。（3）矿产资源预测判别系统的构建从软件工程的角度出发，在对目标系统进行详细的需求分析以及全面的系统设计的基础上，使用 ArcEngine10.1 组件开发包、Oracle 10g Release 2 数据库、ArcSDE 空间数据库引擎以及面向对象语言 C#和 MATLAB 程序解释语言的混合编程语言，开发了具有预测过程自动化与预测结果最优化功能的矿产资源预测判别系统，为用户提供一个界面友好，操作简单的定量预测工作平台。在此基础上，结合矿产资源定量预测的工作流程，探讨了矿产资源预测判别系统在哈图-包古图地区金矿预测工作中的应用，用于验证系统的可行性与实用性。 |
| Other Abstract | Under the background of information age, to overcome the constraint byincreasingly prominent contradiction between per capita hold of resources and resourcereserve, new metallogenic theories and key technologies must be innovated, and newmethods or technologies of mineral resources prediction must be explored. With therapid development of high and new technology, the theories and methods of mineralresources prediction are greatly spring up. Each prediction method and evaluationmethod have its own advantages and disadvantages. Considering the lacking of auniform rule to choose the most suitable method, to explore quantitative predictiontechnique for mineral resources and discriminated technique for mineral resourcesprediction results, an integrated and automated system is proposed to promote theability of mineral resources predication. The thesis studies mathematical theories ofprediction methods and evaluation methods, and combines programming techniquesand mathematical models through computer science. Then GIS technology, databasetechnology and visualization technology are integrated, a mineral resources predictionand discrimination system with prediction process automation and prediction resultsoptimization has been realized, which provides a new technology for mineral resourcesprediction and assessment. The main progress obtained by this thesis includes:(1) Selection of Quantitative Prediction Methods for Mineral ResourcesBasic theories and frontier techniques in Bibliometrics were introduced. Naturallanguage processing and text mining techniques extracted 24 quantitative methods formineral resource prediction. Scientific structures and hot topics and the researchfrontier of quantitative prediction methods for mineral resources were further discussedand analyzed by the co-occurrence analysis and biplot analysis, respectively. Eventually,weights of evidence, fuzzy logic, neural networks, and support vector machine wereselected.(2) The Development of Key Technologies of Mineral Resources Prediction andDiscrimination SystemWeights of evidence, fuzzy logic, BP neural networks, and support vector machineintegrated in mineral resources quantitative prediction technique was studied in detail.Programming techniques and mathematical models were combined and key issues inautomating prediction processes were solved, including conditional independence test, fuzzy inference network, parameter settings of neural networks and cross validation.The conversion of non-automated quantitative mineral resource prediction processesinto automated ones was preliminarily explored.Evaluation methods of prediction results were comprehensively analyzed topropose synthetic discrimination function based on evaluation methods. It can quickly,accurately and objectively discriminate prediction results. Taking ROC and P-A plot asexamples, the specific expression of synthetic discriminant function was confirmed.(3) Construction of a Mineral Resources Prediction and Discrimination SystemThrough software engineering, detailed analyses on requirements of the targetsystem and comprehensive system design, ArcEngine10.1 component development kit,Oracle Database 10g Release 2, ArcSDE spatial database engine, and a hybridprogramming language for the target language C# and MATLAB interpreted languagewere used to develop a mineral resources prediction and discrimination system withprediction process automation and prediction results optimization. It aims to provide auser-friendly and easy-to-use working platform for quantitative prediction. On the basisof the workflow of quantitative prediction of mineral resources, the application of themineral resources prediction and discrimination system in the gold mine assessment inthe Hatu-Baogutu District was investigated. This verifies the feasibility and practicalityof the mineral resources prediction and discrimination system. |
| Subject Area | 地图学与地理信息系统 |
| Language | 中文 |
| Document Type | 学位论文 |
| Identifier | http://ir.xjlas.org/handle/365004/14924 |
| Collection | 研究系统_荒漠环境研究室 |
| 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 | 杜茜诗慧. 矿产资源预测判别系统关键技术的研发与应用[D]. 新疆乌鲁木齐. 中国科学院大学,2018. |
| 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