李诗, 陈建平, 向杰, 张志平, 张烨. 基于AlexNet网络的二维找矿预测——以松桃—花垣地区沉积型锰矿为例[J]. 地质通报, 2019, 38(12): 2022-2032.
    引用本文: 李诗, 陈建平, 向杰, 张志平, 张烨. 基于AlexNet网络的二维找矿预测——以松桃—花垣地区沉积型锰矿为例[J]. 地质通报, 2019, 38(12): 2022-2032.
    LI Shi, CHEN Jianping, XIANG Jie, ZHANG Zhiping, ZHANG Ye. Two-dimensional prospecting prediction based on AlexNet network: A case study of sedimentary Mn deposits in Songtao-Huayuan area[J]. Geological Bulletin of China, 2019, 38(12): 2022-2032.
    Citation: LI Shi, CHEN Jianping, XIANG Jie, ZHANG Zhiping, ZHANG Ye. Two-dimensional prospecting prediction based on AlexNet network: A case study of sedimentary Mn deposits in Songtao-Huayuan area[J]. Geological Bulletin of China, 2019, 38(12): 2022-2032.

    基于AlexNet网络的二维找矿预测——以松桃—花垣地区沉积型锰矿为例

    Two-dimensional prospecting prediction based on AlexNet network: A case study of sedimentary Mn deposits in Songtao-Huayuan area

    • 摘要: 在大数据的时代背景下,地质大数据逐渐趋于复杂化的模式与其间的空间关联性为基于机器学习算法的矿产资源定量预测带来了更大的挑战。利用深度卷积网络算法优异的分析性能来提取不同成矿条件下多种二维要素图层的空间分布特征与关联性是一项非常有意义的探索性实验。以松桃-花垣地区沉积型锰矿为例,利用深度卷积神经网络模型AlexNet挖掘Mn元素、沉积相、大塘坡组出露、断裂及水系的空间分布与锰矿矿床的就位空间的耦合相关性,以及不同的控矿要素之间的相关性,以此训练出二维矿产预测分类模型。经过训练后,可以得到验证准确率88.89%,召回率为66.67%,损失值0.08的深度卷积神经网络分类模型。应用该模型对未知区进行二维成矿预测,共圈定出91、96、154、184号4个成矿远景区,其中91号和154号的区域含矿概率为1,96号含矿概率为0.5。由此可见,预测区具有很大概率存在尚未发现的矿床。

       

      Abstract: There are many challenges in the task of predicting ore deposits from big data repositories. The data are inherently complex and have great significance to the intervenient spatial relevance of deposits. The characteristics of the data make it difficult to use machine learning algorithms for the quantitative prediction of mineral resources. There are considerable interest and value in extracting spatial distribution characteristics from two-dimensional ore-controlling factors'layers under different metallogenic conditions. In this paper, the authors conducted such an analysis by using a Deep Convolutional Neural Network (D-CNN) algorithm named AlexNet. Training on the two-dimensional (2-d) mineral prediction and classification model was performed using data from the Songtao-Huayuan sedimentary manganese deposit. The authors investigated the coupling correlation between the spatial distribution of manganese element, sedimentary facies, outcrop of Datangpo Formation, faults, water system and the areas where manganese orebodies are present, as well as the correlation between different ore-controlling factors by employing the AlexNet networks. After training, the deep convolutional neural network classification model with the verification accuracy of 88.89%, recall of 66.67% and loss value of 0.08 could be obtained. By applying this model to unknown areas for two-dimensional metallogenic prediction, four metallogenic prospective areas. i.e., No. 91, No. 96, No. 154 and No. 184, were delineated, in which the ore potential probability of No. 91 regional ore-bearing probability and No. 154 prospective area is 1, and that of No. 96 is 0.5, suggesting that the probability of existence of undiscovered deposits in prediction areas is large.

       

    /

    返回文章
    返回