管窥人工智能与大数据地球科学研究新进展  

Overview:A glimpse of the latest advances in artificial intelligence and big data geoscience research

作　　者：周永章 肖凡[1,2,3] ZHOU Yongzhang;XIAO Fan(School of Earth Sciences and Engineering,Sun Yat-sen University,Zhuhai 519000,China;Center for Earth Environment&Resources,Sun Yat-sen University,Zhuhai 519000,China;Guangdong Provincial Key Laboratory of Mineral Resources and Geological Processes,Sun Yat-sen University,Zhuhai 519000,China)

机构地区：[1]中山大学地球科学与工程学院,广东珠海519000 [2]中山大学地球环境与资源研究中心,广东珠海519000 [3]中山大学广东省地质过程与矿产资源探查重点实验室,广东珠海519000

出　　处：《地学前缘》2024年第4期1-6,共6页Earth Science Frontiers

基　　金：国家重点研发计划项目(2022YFF0801201);国家自然科学基金联合基金重点项目(U1911202)。

摘　　要：本期是《地学前缘》组织出版的“人工智能与大数据地质”主题专辑。它由17篇学术论文组成,涵盖了知识图谱、基于深度学习的图像识别、非结构化地质信息的机器可读表达、图形大数据与社区发现、关联规则算法、三维地质模拟与成矿预测、物联网与在线监测系统等不同主题,提供了极其有价值的应用场景和研究案例,在一定程度上反映了中国人工智能与大数据地球科学领域研究的最新进展,值得同行关注。This special issue titled“Artificial Intelligence and Big Data Geoscience”consists of 17 papers covering topics such as knowledge graphs,deep learning-based image recognition,machine-readable expression of unstructured geological information,big graph data and community detection,association rule algorithms,3D geological simulation and mineral prospecting,and the Internet of Things and online monitoring systems.A progressive multi-granularity training deep learning method is proposed for mineral image identification;the model achieves 86.5%accuracy on a commonly used dataset comprising 36 mineral types,increasing the accuracy of mineral identification.Knowledge related to porphyry copper ore in the Qinzhou-Hangzhou mineralization belt,South China,is collected using both primary and literature data sources,and Natural Language Processing(NLP)techniques are used to semantically correlate and reason over the knowledge graph,enabling automated knowledge extraction and reasoning.The association rule algorithm is used to analyze the correlation between trace elements and gold mineralization in major Carlin-type gold deposits in the“Golden Triangle”region of Yunnan-Guizhou-Guangxi provinces,China,and combined with the migration and enrichment law of elements to analyze the genetic mechanism of deposits.By builing a quantitative prospecting indicator method based on association rule algorithm,this study provides new ideas for establishing quantitative prospecting indicators for other types of deposits.In study of machine-readable expression of unstructured geological information and intelligent prediction of mineralization associated anomaly areas in Pangxidong District,western Guangdong,China,unstructured geological information such as stratigraphy,lithology and faults is processed by machine-readable conversion,and two machine learning algorithms-namely,One-Class Support Vector Machine and Auto-Encoder network-are applied to mine the geochemical test data of the stream sediment as well as the comprehensive geol

关 键 词：知识图谱 深度学习 图像自动识别 非结构地质信息 社区发现 大数据挖掘 三维地质建模 物联网标识 

分 类 号：P59[天文地球—地球化学] P628[天文地球—地质学] TP18[自动化与计算机技术—控制理论与控制工程]