| 彭雪峰,姜丽萍,蔡图,陈瑶.基于 SSA-CNN的多源信息集成驱动找矿模型——以胶莱盆地南缘七宝山陆相火山热液型矿床为例[J].矿产勘查,2026,17(1):175-184 |
| 基于 SSA-CNN的多源信息集成驱动找矿模型——以胶莱盆地南缘七宝山陆相火山热液型矿床为例 |
| Multi-source information integration driven prospecting model based on SSA-CNN: A case study of Qibaoshan continental volcanic hydrothermal deposit in the southern margin of Jiaolai Basin |
| 投稿时间：2025-02-18 |
| DOI：10.20008/j.kckc.202601014 |
| 中文关键词: 卷积神经网络 多源信息集成 找矿模型 胶莱盆地南缘 |
| 英文关键词: convolutional neural network multi-source information integration prospecting model south mar-gin of Jiaolai Basin |
| 基金项目:本文受山东省省级地质勘查项目“山东省1∶5万枳沟、诸城幅区域矿产地质调查”(鲁勘字(2019)18号资助。 |
|
| 摘要点击次数: 232 |
| 全文下载次数: 96 |
| 中文摘要: |
| 随着全球工业化进程的推进，矿产资源需求持续增长，深部矿床及复杂地质背景下的勘探工作面临日益严峻的挑战，传统找矿方法在处理复杂地质数据时的局限性愈发显著。本研究基于胶莱盆地南缘七宝山陆相火山热液型矿床，提出基于深度学习的多源信息集成驱动找矿模型。通过融合遥感、地球物理和地球化学等多源数据，利用麻雀搜索算法优化的卷积神经网络(SSA-CNN)自动提取数据空间特征，构建高精度的光谱-物探-化探多源融合找矿预测模型，提升了高维数据利用率和模型精度。研究结果显示，模型预测的找矿靶区与已知矿体和矿脉的重合度达 86.4%，并通过槽探工程验证了模型的可靠性。该研究提升了现有找矿模型在处理复杂地质数据和进行非线性预测时的准确性，为其他地质背景下的矿产勘查提供了可借鉴的技术路线和理论依据。 |
| 英文摘要: |
| With the advancement of the global industrialization process, the demand for mineral resources con-tinues to grow, and the exploration of deep deposits and complex geological backgrounds is facing increasingly se-vere challenges. The limitations of traditional prospecting methods in processing complex geological data are becom-ing more and more significant. Based on the Qibaoshan continental volcanic hydrothermal deposit in the southernmargin of Jiaolai Basin, this paper proposes a deep learning-based multi-source information integration driven pros-pecting model. By integrating remote sensing, geophysical and geochemical multi-source data, the convolutionalneural network (SSA-CNN) optimized by Sparrow search algorithm was used to automatically extract spatial featuresof the data, and a high-precision spectral, geophysical and geochemical multi-source fusion prospecting predictionmodel was constructed, which improved the utilization rate of high-dimensional data and model accuracy. The re-sults show that the coincidence degree of the prospecting target predicted by the model with the known ore body andvein is 86.4%, and the reliability of the model is verified by the trough exploration engineering. This study improvesthe accuracy of the existing prospecting model in processing complex geological data and nonlinear prediction, andprovides a useful technical route and theoretical basis for mineral exploration in other geological backgrounds. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |