金属矿山深部资源三维智能预测理论与方法

Theory and methods for three-dimensional intelligent prospectivity mapping of deep resources in metal mines

当前,中国主要矿产大多面临后备资源不足问题,在大中型矿山深部寻找可接替资源已成为保障主要矿产的战略措施。然而,金属矿山深部找矿预测面临着矿床深部结构不清、深部控矿规律隐蔽、深部矿体空间定位难度大等关键问题,亟须建立适应矿山真三维空间要求的矿产资源预测与评价新理论并取得关键技术突破。针对这些问题,自20世纪80年代开始,笔者通过持续的探索和研究创新,先后提出隐伏矿体立体定量预测、深部资源三维可视化预测,最终提出并建立深部资源三维智能预测理论与方法,实现了矿床深部三维结构重建的自动化与精细化、深部控矿规律表征的定量化与透明化、深部成矿空间矿体定位的精准化与智能化。该理论与方法主要包括成矿系统分析与找矿概念模型构建、矿床深部结构贝叶斯数据同化三维精细重建、矿床深部地质结构三维几何-物质分析与成矿信息提取、深部矿体空间定位规律深度学习与三维预测等重要方法与关键技术。该理论与方法先后在山东胶西北金矿集区、甘肃金川铜镍矿等地取得成功应用,在矿区深部探获厚度大、高品位矿体,取得深部找矿突破。

At present,China is facing a severe shortage of backup mineral resources.Finding replaceable resources in the deep parts of large and medium-sized mines has become a strategic measure to ensure major mineral supplies.However,the deep mineral prospectivity mapping faces key challenges such as unclear deep orebody structures,hidden deep ore-controlling patterns,and significant difficulties in spatial positioning of deep orebodies.There is an urgent need to establish new theories for mineral prospectivity mapping and evaluation that meet the requirements of true three-dimensional(3D)spatial modeling in mining,and to achieve key technological breakthroughs.To address these issues,starting from the 1980s,the author has proposed and developed innova⁃tive approaches through continuous exploration and research.These include the stereoscopic quantitative predic⁃tion of concealed orebodies and the visualization of deep 3D mineral prospectivity mapping.Ultimately,the theory and methods for 3D intelligent prospectivity mapping of deep metal mine resources were proposed and established.These methods have automated and refined the 3D reconstruction of deep orebody structures,quantified and made transparent the representation of deep ore-controlling patterns,and enabled precise and intelligent spatial localiza⁃tion of deep mineralization zones.The theory and methods for deep 3D intelligent mineral prospectivity mapping mainly include:analysis of metallogenic systems and construction of ore-finding conceptual models,Bayesian data assimilation for 3D detailed reconstruction of deep orebody structures,3D geometric-material analysis of deep geological structures and extraction of metallogenic information,deep learning and 3D prospectivity mapping of spatial localization rules for deep orebodies.These theories and methods have been successfully applied in areas such as the Jiaodong Peninsula in Shandong and the Jinchuan copper-nickel mine in Gansu,achieving significant breakthroughs in deep prospecting by discovering thick,high-grade orebodies in deep mining areas.