卷积神经网络在山东金矿勘查预测中的应用

doi:10.11720/wtyht.2023.1613

摘要
图/表
参考文献
相关文章
Metrics

全文: PDF(5261 KB) HTML
输出: BibTeX | EndNote (RIS)

摘要

大数据和人工智能技术在矿产资源预测方面的应用已得到快速发展,但基于卷积神经网络机器学习技术的应用仍处于探讨和试验阶段,我国的矿产资源勘查预测实用化的例子和成果较少。针对上述问题,提出了将卷积神经网络应用到金矿勘查中,根据山东省某地金矿成矿区域的3×10⁴km²范围内的地质、矿产、地球物理、地球化学等实测专业数据资料,进行了2 000轮卷积神经网络的训练,最终得到了准确率0.95、损失率0.11的一维卷积神经网络模型。将这套卷积神经网络用于山东省其他未知地区,进行金矿床分布位置(勘查靶区)的预测试验,得到了较好的结果。

	服务

	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	郑孝诚
	张明华
	任伟

关键词 ：卷积神经网络, 机器学习, 金矿勘探

Abstract：

Rapid progress has been made in the application of big data and artificial intelligence technology in the prediction of mineral resources. However, the application of machine learning technology based on convolutional neural networks remains in the exploration and experimental stages, with few practical examples and accomplishments achieved in the exploration and prediction of mineral resources in China. This study proposed applying convolutional neural networks to the exploration of gold deposits. Specifically, a neural network was trained for 2000 rounds using measured geological, mineral, geophysical, and geochemical data collected from a mineralization region covering an area of 3×10⁴ km² in a gold deposit in Shandong Province. Consequently, a 1D convolutional neural network model with accuracy of 0.95 and a loss rate of 0.11 was obtained. This model was employed to predict the distribution locations of gold deposits (exploration target areas) in other unknown areas in Shandong Province, yielding encouraging outcomes.

Key words： convolutional neural network machine learning gold exploration

收稿日期: 2022-12-22 修回日期: 2023-03-30 出版日期: 2023-12-20

P631

基金资助:国家自然科学基金项目“图文数一体化的地学知识图谱构建与应用”(42050105)

通讯作者: 任伟(1985-),男,高级工程师,主要从事地球物理与地质信息研究工作。Email:renwei@mail.cgs.gov.cn

作者简介: 郑孝诚(1995-),男,硕士研究生,研究方向为地质工程。Email:928072533@qq.com

引用本文:

郑孝诚, 张明华, 任伟. 卷积神经网络在山东金矿勘查预测中的应用[J]. 物探与化探, 2023, 47(6): 1433-1440.
ZHENG Xiao-Cheng, ZHANG Ming-Hua, REN-Wei . Application of convolution neural networks in gold exploration and prediction in Shandong Province. Geophysical and Geochemical Exploration, 2023, 47(6): 1433-1440.

链接本文:

https://www.wutanyuhuatan.com/CN/10.11720/wtyht.2023.1613 或 https://www.wutanyuhuatan.com/CN/Y2023/V47/I6/1433

Fig.1 研究区地质概况

Fig.2 研究流程

Table 1 不同岩性和地质构造标签定义

Fig.3 训练集的点位区域

Fig.4 训练后的准确率函数与损失函数

Fig.5 验证集点位分布

Fig.6 验证集的输出等概率图

Fig.7 实验测试集数据选取的点位分布

Fig.8 测试集的输出等概率图

[1]	赵鹏大. 成矿定量预测与深部找矿[J]. 地学前缘, 2007, 14(5):1-10.
[1]	Zhao P D. Quantitative mineral prediction and deep mineral exploration[J]. Earth Science Frontiers, 2007, 14(5):1-10.
[2]	袁峰, 李晓晖, 张明明, 等. 隐伏矿体三维综合信息成矿预测方法[J]. 地质学报, 2014, 88(4):630-643.
[2]	Yuan F, Li X H, Zhang M M, et al. A method for three-dimensional comprehensive information mineralization prediction of concealed ore bodies[J]. Journal of Geology, 2014, 88(4):630-643.
[3]	Harris D P. Undiscovered uranium resources and potential supply[C]// Workshop on Concepts of Uranium Resources and Producibility. Washington: Board Mineral Energy Resources,National Academy of Sciences, 1978:51-81.
[4]	Singer D A. Basic concepts in three-part quantitative assessments of undiscovered mineral resources[J]. Nonrenewable Resources, 1993, 2 (2):69-81. doi: 10.1007/BF02272804
[5]	赵鹏大. “三联式”资源定量预测与评价——数字找矿理论与实践探讨[J]. 地球科学—中国地质大学学报, 2002, 27(5):482-489.
[5]	Zhao P D. Quantitative prediction and evaluation of "Three Linked" resources:A discussion on digital prospecting theory and practice[J]. Earth Science-Journal of China University of Geosciences, 2002, 27(5):482-489.
[6]	赵鹏大, 胡旺亮. 地质异常理论与矿产预测[J]. 新疆地质, 1992, 10(2):93-100.
[6]	Zhao P D, Hu W L. Geological anomaly theory and mineral prediction[J]. Xinjiang Geology, 1992, 10(2):93-100.
[7]	王世称. 综合信息矿产预测理论与方法体系新进展[J]. 地质通报, 2010, 29(10):1399-1403.
[7]	Wang S C. New progress in comprehensive information mineral prediction theory and methodology[J]. Geological Bulletin, 2010, 29(10):1399-1403.
[8]	张勇, 马晓东, 李彦婧, 等. 深度学习在南川页岩气含气量预测中的应用[J]. 物探与化探, 2021, 45 (3):569-575.
[8]	Zhang Y, Ma X D, Li Y J, et al. The application of deep learning to the shale gas content prediction in Nanchuan(South Sichuan)[J]. Geophysical and Geochemical Exploration, 2021, 45 (3):569-575.
[9]	成秋明. 非线性成矿预测理论:多重分形奇异性—广义自相似性—分形谱系模型与方法[J]. 地球科学—中国地质大学学报, 2006, 31(3):337-348.
[9]	Cheng Q M. Nonlinear ore-forming prediction theory:Multifractal singularity-generalized self similarity-fractal lineage model and method[J]. Earth Science-Journal of China University of Geosciences, 2006, 31(3):337-348.
[10]	周永章, 黎培兴, 王树功, 等. 矿床大数据及智能矿床模型研究背景与进展[J]. 矿物岩石地球化学通报, 2017, 36(2):327-331,344.
[10]	Zhou Y Z, Li P X, Wang S G, et al. Research background and progress of big data and intelligent deposit models for mineral deposits[J]. Bulletin of Mineral and Rock Geochemistry, 2017, 36(2):327-331,344.
[11]	刘艳鹏, 朱立新, 周永章. 卷积神经网络及其在矿床找矿预测中的应用—以安徽省兆吉口铅锌矿床为例[J]. 岩石学报, 2018, 34(11):3217-3224.
[11]	Liu Y P, Zhu L X, Zhou Y Z. Convolutional neural networks and their application in ore prospecting and prediction:A case study of the Zhaojikou lead-zinc deposit in Anhui Province[J]. Acta Petrologica Sinica, 2018, 34(11):3217-3224.

[1]	王康, 刘彩云, 熊杰, 王永昌, 胡焕发, 康佳帅. 基于全卷积残差收缩网络的地震波阻抗反演[J]. 物探与化探, 2023, 47(6): 1538-1546.
[2]	周慧, 孙成禹, 刘英昌, 蔡瑞乾. 基于DC-UNet卷积神经网络的强噪声压制方法[J]. 物探与化探, 2023, 47(5): 1288-1297.
[3]	宫明旭, 白利舸, 曾昭发, 吴丰收. 基于机器学习松辽盆地大地热流计算与特征分析[J]. 物探与化探, 2023, 47(3): 766-774.
[4]	梁立锋, 刘秀娟, 张宏兵, 陈程浩, 陈锦华. 超参数对GRU-CNN混合深度学习弹性阻抗反演影响研究[J]. 物探与化探, 2021, 45(1): 133-139.
[5]	刘仕友, 宋炜, 应明雄, 孙万元, 汪锐. 基于波形特征向量的凝聚层次聚类地震相分析[J]. 物探与化探, 2020, 44(2): 339-349.
[6]	王逸宸, 柳林涛, 许厚泽. 基于卷积神经网络识别重力异常体[J]. 物探与化探, 2020, 44(2): 394-400.

Viewed

Full text

Abstract

Cited

Shared

Discussed