Research on Integrated Analysis of Geological and Geophysical Data and 3D Mineralization Prediction Based on Machine Learning

A machine learning-based method for mineralization prediction is proposed, leveraging a 3D geological-geophysical model, aiming to achieve precise delineation of three-dimensional prospective mineral exploration targets. This approach integrates multi-source geophysical property parameters, such as density, magnetic susceptibility, and resistivity, with regional geological settings, ore deposit characteristics, and drilling data. A mineralization prediction model is established based on machine learning algorithms to address parameter overlap and inherent geological ambiguity. Algorithms such as Random Forest and Support Vector Machines are employed to achieve nonlinear mapping and efficient classification of the data, while grid search is used to optimize model parameters, leading to notable improvements in prediction accuracy and reliability. Model performance is evaluated through cross-validation, demonstrating its applicability. Applied to the Duobaoshan ore district in Heilongjiang Province, China, a well-known mineralized region, this method successfully delineated multiple 3D prospective exploration targets, showcasing its potential in the integrated analysis and 3D modeling of geological and geophysical data. This study provides new insights and technical support for mineralization prediction under complex geological conditions.

Keywords: Multi-source geological-geophysical data; 3D modeling; Machine learning; 3D targeting