Relevance. Coal remains a crucial fossil fuel source globally, significantly contributing to economic and industrial development. Efficient coal extraction relies heavily on accurate identification and evaluation of coal deposits, where geophysical methods, particularly well-logging, play a pivotal role. Aim. To enhance the accuracy and efficiency of determining coal seam depth and thickness by applying advanced machine learning techniques to well-logging geophysical data. This approach seeks to optimize coal exploration by automating data analysis, reducing human error, and improving the precision of subsurface assessments critical for efficient resource extraction. Objects. Well-logging geophysical data collected from multiple boreholes in a coal mine in Quang Ninh province, Vietnam. This region, a hub for coal production, provides a rich dataset for testing and refining machine learning models tailored to real-world mining conditions. Methods. A variety of machine learning algorithms, including regression and classification techniques, were utilized to process and analyze the well-logging data. These models were trained and validated using labeled datasets derived from borehole measurements, enabling accurate predictions of coal seam depth and thickness with enhanced precision and reliability. Results. The machine learning models achieved an impressive predictive accuracy of up to 89% when compared to actual sample data. These findings demonstrate the transformative potential of integrating artificial intelligence into geophysical analysis. By offering improved reliability, time efficiency, and accuracy in coal seam characterization, this approach supports more sustainable and cost-effective coal extraction practices, paving the way for broader adoption in the mining industry.

For citation: Vu Hong Duong, Nguyen Tien Hung, Nguyen Minh Hoa, Nguyen The Vinh. Machine learning-based analysis of borehole geophysical logs in coal mining. Bulletin of the Tomsk Polytechnic University. Geo Assets Engineering, 2026, vol. 337, no. 6, pp. 44-55. https://doi.org/10.18799/24131830/2026/6/5224