Szczegóły publikacji
Opis bibliograficzny
Dynamic management of ore body bedding models based on machine learning and terrain grid reconstruction / Dmytro Malashkevych, Vladyslav Ruskykh, Marek DUDEK, Dariusz SALA, Yuliya PAZYNICH // W: Proceedings of the DIM-ESEE Conference [Dokument elektroniczny] : Dubrovnik, October 15th – 17th, 2025 / eds. Dalibor Kuhinek, [et al.]. — Wersja do Windows. — Dane tekstowe. — Zagreb : University of Zagreb. Faculty of Mining, Geology and Petroleum Engineering, 2025. — ( Proceedings of the DIM-ESEE Conference ; ISSN 3102-3517 ; vol. 1 ). — S. MM1–MM7. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. MM6, Abstr. — Y. Pazynich - dod. afiliacja: Dnipro University of Technology, Dnipro, Ukraine
Autorzy (5)
- Malashkevych Dmytro
- Ruskykh Vladyslav
- AGHDudek Marek
- AGHSala Dariusz
- AGHPazynich Yuliya
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 166744 |
|---|---|
| Data dodania do BaDAP | 2026-03-31 |
| Tekst źródłowy | URL |
| Rok publikacji | 2025 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Creative Commons | |
| Czasopismo/seria | Proceedings of the DIM-ESEE Conference |
Abstract
Accurate three-dimensional modeling of ore bodies is fundamental for effective mine planning, reserve estimation, and production management. However, due to the limitations of exploration methods, geological survey data are often incomplete, making it necessary to continuously update deposit models as new information becomes available. This study presents a methodology for the dynamic updating of ore body models that integrates geological data analysis with machine learning algorithms and mesh reconstruction techniques. The approach was applied to the Southern Bilozerka iron ore deposit, developed by PJSC Zaporizhzhia Iron Ore Plant. The methodology combines Linear Regression and Random Forest algorithms to enhance predictive accuracy. Linear Regression provides interpolation functions describing the general ore body geometry, while Random Forest improves the robustness of forecasts by reducing errors under heterogeneous geological conditions. Using Python-based tools in the Google Colab environment, predictive points of ore body distribution were calculated and then incorporated into wireframe and block models. This allowed not only the refinement of ore body geometry but also reliable reserve estimation, which for the studied level reached approximately 22 million tonnes. Furthermore, the constructed models formed a basis for calendar-based planning of development and stoping operations. The results confirmed that combining regression and ensemble approaches is an effective solution for dynamically updating geological models of ore bodies. The methodology ensures the integration of mine surveying and exploration data into digital models, improving the reliability of long-term planning and mine design.
