基于贝叶斯框架的深部成矿构造推断及不确定性研究

Deep Metallogenic Structure Inference and Uncertainty Study Based on Bayesian Framework

摘要: 矿产资源评价目前正逐步向三维定量预测方向发展，然而，随着预测深度的增加，由于深部探测数据减少，深部三维模型的不确定性也逐渐增长，因此如何减少不确定性以及如何对不确定性进行很好的度量是目前三维定量预测所研究的重要问题。本文提出一种基于贝叶斯框架的深部成矿构造推断及不确定性度量方法，利用该方法对大尹格庄金矿床成矿构造招平断裂面深部结构进行贝叶斯推断及不确定度量，最终基于推断得到的断裂面形态后验概率，得出深部成矿构造不确定性可视化结果。研究表明，通过该方法得到的不确定性可视化模型与地质认识相符，可为后续的深部成矿构造三维模型修正提供引导。

Abstract: Mineral resources evaluation is now gradually moving towards 3D quantitative prediction. However, with the increase of the predicted depth, the uncertainty of the metallogenic structure located in the deep due to its scarce data has always existed. Therefore, how to reduce the uncertainty and how to measure uncertainty is a hot issue to study the 3D quantitative prediction. This paper mainly uses the Bayesian theory to construct the Bayesian model, and based on this model, the Bayesian inference is made for the deep metallogenic structure of the Dayingezhuang mining area, and finally obtains a visual model based on the deep metallogenic structure. Research shows that the uncertainty visualization model obtained through this example is consistent with geological research, and can provide guidance for subsequent iterative inference of 3D models of deep metallogenic structures.