神经网络在砂岩型铀矿测井中的岩性识别应用
Application of Neural Network in Lithology Identification in Sandstone-type Uranium Deposits
陈维政1
作者信息
- 1. 核工业二一六大队,乌鲁木齐 830000
- 折叠
摘要
对测井数据进行岩性划分是砂岩型铀矿勘探的重要工作.采用人工神经网络对砂岩型铀矿的测井数据进行岩性识别,通过两种方式计算隐藏层神经元的数量,使用GridSearchCV优选出最优隐藏层神经元个数及学习率,通过对实际数据进行识别,模型可以在200次迭代内达到收敛.混淆矩阵结果表明,对可渗透的砂岩识别准确率达80％,对不可渗透的粉砂岩、泥质粉砂岩识别准确率达到60％.
Abstract
The lithology division of logging records is an important work in the exploration of sandstone-type uranium deposits.Artificial neural network is used to identify lithology from log data of sandstone-type uranium deposits.Two methods are used to calculate the number of hidden layer neurons,and GridSearchCV is used to optimize the number of hidden layer neurons and the learning rate.The results of the confusion matrix show that the recognition accuracy of permeable sandstone can reach 80％,and the recognition accuracy of impenetrable siltstone and muddy siltstone can reach 60％.
关键词
岩性识别/神经网络/混淆矩阵/机器学习Key words
Lithology identification/Neural network/Confusion matrix/Machine learning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据