Classification Identification And Potential Evaluation Of Magmatic Hydrothermal Deposits Based On Machine Learning

With the rapid development of information technology and artificial intelligence,the emergence of big data and machine learning has changed the way of thinking and research methods in geological and mineral exploration research,bringing new opportunities and challenges for the processing and application of massive geological data.The magmatic hydrothermal activity can form a variety of different types of metal minerals,and the oreforming magmatic magmatic magmatic magmatic magmatic magmatic magmatic magmatic magmatic magmatic hydrothermal activity can form a variety of different types of metal minerals,so it is important to effectively distinguish between different types of magmatic hydrothermal deposits;South China,as an important uranium-forming province in China,has widely developed granite bodies,but after years of continuous mining,it has become increasingly difficult to find ore.Therefore,how to use the effective combination of geochemical data and big data machine learning methods to determine different deposit types and evaluate the potential of minerals is of great importance to guide mineral exploration.In this paper,machine learning and data mining methods are applied to classify magmatic hydrothermal deposits,mine indicator elements,and evaluate the uranium potential of granitic types in South China based on rock geochemical data with the guidance of machine learning and big data ideas.The main achievements are as follows:(1)Classification of magmatic hydrothermal deposits: Random forest,support vector machine,and artificial neural network models were applied to classify five magmatic hydrothermal deposits,among which the accuracy of the random forest model was 93%,the support vector machine model was 89%,and the artificial neural network was 83%.Since there was no significant change in the results of the random forest model after adjusting the parameters,this paper considers the support vector machine model as the three The support vector machine model is considered the most suitable model among the three models in this study.(2)Data mining of magmatic hydrothermal deposits: Using the ratio of geochemical elements,confidence ellipse discriminant maps are constructed to fully mine the indicative role of geochemical elements,and a series of new discriminant maps are obtained.The results that the effective combination of machine learning,big data,and geochemical data can be used to effectively discriminate their deposit types and explore the indicative elements in them.(3)Evaluation of granite-type uranium potential in South China: We systematically standardized the collection of geochemical elemental content data on South China granites from previous published literature,and established random forest algorithm and K-nearest neighbor algorithm classification models,respectively.The recall,ROC curve,and AUC values of the random forest model were better than those of the K-nearest neighbor model,with an accuracy of 93%.Using the random forest model created above to predict the Nine Peaks,Red Mountain,and Tea Mountain rock bodies,the Red Mountain rock body and the Tea Mountain rock body have a higher probability of containing ore,while the Nine Peaks rock body has a lower probability of containing ore.The model can be used as an auxiliary tool for geologists.