摘要
在地浸砂岩型铀矿勘查过程中,砂体渗透系数是一项重要的水文地质参数,也是评价地浸开采有效性的重要指标,而砂体渗透系数预测及其空间分布特征是铀矿勘查最为困难的问题之一。机器学习算法具有操作简单及泛化能力强的特点,在各领域得到了广泛应用,将机器学习算法引入到铀矿地质勘查中能够快速有效地解决测井数据与砂体渗透系数之间复杂的非线性问题。文章采用随机森林回归算法预测洪海沟铀矿床砂体渗透系数,并将其与非线性回归、支持向量机回归预测结果进行对比评价,结果表明,随机森林回归算法具有拟合优度高、操作简单的特点。该方法预测出的砂体渗透系数空间展布方向与地下水渗流方向一致,并密切影响层间氧化带及铀矿化的发育,具有一定的推广意义和经济价值。
In the process of in-situ leaching of sandstone-type uranium deposit,the sand permeability coefficient is an important hydrogeological parameter which plays an important role to evaluate the effectiveness of ground leaching mining.However,the prediction of sand permeability coefficient and its spatial distribution characteristics is one of the most difficult problems in uranium exploration.Due to the simple operation and strong generalization ability,machine learning algorithms have been widely used in various fields.In this paper,the machine learning was tested to quickly and effectively solve the complex nonlinear problems between logging data and sand permeability coefficients,the random forest regression algorithm was used to predict the permeability coefficient of sand bodies of uranium deposits in the Flood Trench,and compared with the prediction results of support vector machine regression.The comparison results showed that the random forest regression algorithm was of high goodness-of-fit and simple operation.The predicted spatial distribution direction of the permeability coefficient of sand body is consistent with the direction of groundwater seepage,which closely affects the development of interlayer oxidation zone,therefore the proposed prediction method has economic and reference value.
作者
刘富强
陈晓冬
李盛富
张福衡
LIU Fuqiang;CHEN Xiaodong;LI Shengfu;ZHANG Fuheng(Geologic Party No.216,CNNC,Urumqi,Xinjiang 830011,China)
出处
《铀矿地质》
CAS
CSCD
2023年第4期653-661,共9页
Uranium Geology
基金
中国核工业地质局项目“新疆伊犁盆地及周边地区砂岩型铀矿资源调查评价与勘查”(编号:202205)资助。
关键词
随机森林回归
机器学习
渗透系数预测
random forest regression
machine learning
prediction of permeability coefficient
