Full text loading...
-
Predicting Sonic Velocities from Borehole Data Using Neural Networks: Case Study from the Ludvika Mine, Sweden
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, NSG 2025: 6th Conference on Geophysics for Mineral Exploration and Mining, Sep 2025, Volume 2025, p.1 - 5
Abstract
Summary
Accurate prediction of subsurface physical properties is essential in geophysical exploration, particularly in areas with limited or inaccessible borehole measurements. This study applies neural networks to predict sonic (P-wave) velocities in the Blötberget iron-oxide deposit in Ludvika, central Sweden. The dataset comprises lithological and geophysical logs from four key boreholes selected from an archive of over 400, incorporating parameters such as density, magnetic susceptibility, resistivity, gamma radiation, and lithological descriptions. Two NN models were developed: an Accurate model tuned for precision and a Generic model designed for generalization across varied geological settings. The findings demonstrate the feasibility and value of machine learning-driven property prediction in mineral exploration, supporting more cost-effective, data-informed exploration workflows, particularly in structurally complex and data-limited environments.
© EAGE Publications BV