基于深度学习的深层次矿化信息挖掘与集成

Deep Learning-Based Mining and Integration of Deep-Level Mineralization Information

矿产预测的核心是对地学空间数据进行特征提取与集成融合,当前的研究热点和前沿聚焦于深层次矿化信息特征提取与集成。进入大数据时代,如何基于机器学习开展深层次矿化信息挖掘与集成是当前矿产预测的前沿领域。本文介绍了基于机器学习的矿产预测与评价研究的主要内容,深度学习的基本原理,以及深度学习在地球化学异常识别和多源找矿信息集成融合中的应用。研究结果表明,深度学习可有效识别和提取地球化学异常,并能对地质、地球物理、地球化学等多源地学数据进行特征提取、集成融合及找矿远景区圈定。尽管如此,如何把深度学习与地质约束有机结合,使其既能有效挖掘与集成深层次矿化信息,又符合地质认知,还需要更加深入的研究。

The core of mineral resources prediction and assessment is the feature extraction and integration of various geospatial data. The current research hotspot and frontier focus on the feature extraction and integration of the deep-level mineralization information. In the era of big data,how to carry out deep mining and integration of deep-level mineralization information based on machine learning algorithms is one of the frontier subjects in current mineral exploration. This paper briefly introduces the main research contents of mineral prediction and evaluation based on machine learning algorithms,the basic principle of deep learning algorithms,and the application of deep learning algorithms in the recognition of geochemical anomalies and integration of multi-source prospecting information. The results demonstrate that the deep learning algorithms can effectively identify and extract geochemical anomalies,and can extract characteristics and integrate geological, geophysical, geochemical and other multi-source geoscientific data, and delineate the prospecting area.Nevertheless,how to combine deep learning algorithms with geological constraints so that they can not only effectively excavate and integrate deep-level mineralization information,but also coincide with geological cognition,and more in-depth research is needed.