铀矿冶2025，Vol.44Issue(2)：9-17,37,10.DOI:10.13426/j.cnki.yky.2024.10.09
基于不同机器学习算法的地浸测井数据岩性识别对比
Lithology Identification Comparison on Geophysical Logging Data of In-situ Leaching Uranium Based on Multiple Machine Learning Algorithms
摘要
Abstract
Machine learning algorithms can automatically learn and extract features from a large amount of ge-ological data to achieve fast and accurate lithology identification.In this paper,the logging data of several wells in a sandstone-type uranium deposit in Inner Mongolia were randomly divided into training sets and veri-fication sets according to the ratio of 7∶2.The model structure was adjusted and the hyperparameters were optimized for training.BC1401,BC2802,BC4603 and BC7206 well were used for testing to realize the com-parative analysis of 5 kinds of models,such as random forest,XGBoost,K value proximity algorithm,BP neural network and SMOTE-LSTM algorithm.The results show that SMOTE-LSTM model has the most superior stability and accuracy,with an accuracy of 84.6％.关键词
地浸采铀/机器学习/砂岩型铀矿/岩性识别/测井数据/SMOTE-LSTM模型/XGBoost模型Key words
in-situ leaching of uranium/machine learning/sandstone-type uranium deposit/lithology identification/logging data/SMOTE-LSTM model/XGBoost model分类
矿业与冶金引用本文复制引用
邹玉涵,李沁慈,阙为民,黄亮,吴童盼,杜志明,姜振蛟,陈炫沂..基于不同机器学习算法的地浸测井数据岩性识别对比[J].铀矿冶,2025,44(2):9-17,37,10.基金项目
中核集团青年英才项目(A101-8,基于地浸钻孔数据的采区多源信息智能化提取研究). （A101-8,基于地浸钻孔数据的采区多源信息智能化提取研究）
