%0 Journal Article
%T 基于机器学习的焦作丹河电厂三维地质建模
Three-Dimensional Geological Modeling of Jiaozuo Danhe Power Plant Based on Machine Learning
%A 潘星昊
%A 吴琦
%J Advances in Geosciences
%P 1513-1521
%@ 2163-3975
%D 2024
%I Hans Publishing
%R 10.12677/ag.2024.1412142
%X 在岩体工程评估、地基基础设计及不良地质条件处理中，快速且高效的三维地质建模方法发挥着至关重要的作用。传统地质建模技术往往面临操作难度大、流程繁琐及建模精度低等问题。基于机器学习的方法，将传统的三维地质建模问题转化为对地下空间栅格进行地层属性分类的问题。以焦作丹河电厂异地扩建项目为例，通过比较决策树、随机森林、支持向量机与XGBoost这四种分类算法的效果发现，在无需额外钻孔数据或虚拟钻孔支持的情况下，XGBoost表现出更高的准确率和更优的建模性能。
In the evaluation of rock engineering, foundation design, and the treatment of adverse geological conditions, rapid and efficient three-dimensional (3D) geological modeling methods play a crucial role. Traditional geological modeling techniques often encounter issues such as high operational difficulty, cumbersome procedures, and low modeling accuracy. By employing machine learning methods, the conventional problem of 3D geological modeling is transformed into one of classifying stratigraphic properties within a subsurface grid. Using the expansion project of the Danhe Power Plant in Jiaozuo as a case study, a comparison of the performance of four classification algorithms—decision trees, random forests, support vector machines, and XGBoost—revealed that XGBoost demonstrated higher accuracy and superior modeling performance without the need for additional borehole data or virtual boreholes.
%K 机器学习，
%K 三维地质建模，
%K XGBoost
Machine Learning
%K 3D Geological Modeling
%K XGBoost
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=103940