Application of artificial intelligence algorithm in processing structural geological models

An example of learning a neural network capable of processing and classifying triangulation frameworks of structural geological models is presented. The classification of the frameworks uses the proper calculated vectors of lithological surfaces. The accuracy of the models is 98.2%, which confirms the usability of the machine learning algorithm in classifying and selecting objects at the increased speed of data processing and at the reduced likelihood of errors due to the human factor. After separating the frameworks of tectonic faults and lithological contacts from structural geological models, an additional information on spatial position (average dip angle, dip azimuth) and surface characteristics, such as roughness and waviness, is determined for each framework, which allows using the obtained data to perform kinematic analyses and for solving problems connected with rock movements. The Woodcock coefficients and the eigenvalues of the vectors are used as input data in the model being trained. These characteristics can be used for a refined assessment of the state of a block massif and used to determine the degree of adhesion of tectonic faults, which makes it possible to improve the accuracy of predictive models and subsequently be used to train models that describe the parameters of displacement of the overlying rock mass.

Sergunin M. P., Eremenko V. A. Application of artificial intelligence algorithm in processing structural geological models. MIAB. Mining Inf. Anal. Bull. 2023;(9):56-67. [In Russ]. DOI: 10.25018/0236_1493_2023_9_0_56.

M.P. Sergunin, Head of Department, Center for Geodynamic Safety, Polar Division of MMC Norilsk Nickel, Norilsk, Russia, e-mail: SerguninMP@nornik.ru, ORCID ID: 0000-0002-7774-6826,
V.A. Eremenko, Dr. Sci. (Eng.), Professor of Russian Academy of Sciences, Director of the Research Center for Applied Geomechanics and Convergent Technologies in Mining, Professor at Department of Physical Processes in Mining and Geocontrol, Mining Institute, National University of Science and Technology «MISiS», 119049, Moscow, Russia, e-mail: prof.eremenko@gmail.com, ORCID ID: 0000-0003-1478-6916.

M.P. Sergunin, e-mail: SerguninMP@nornik.ru.

1. Melnichenko I. A., Kozhukhov A. A., Omelchenko D. R., Moseykin V. V. 3D mineral deposit modeling using concepts of blockmodeling and artificial neural networks. MIAB. Mining Inf. Anal. Bull. 2022, no. 10, pp. 5—19. [In Russ]. DOI: 10.25018/0236_1493_2022_10_0_5.

2. Sergunin M. P., Eremenko V.A. Learning of neural network to predict overlying rock mass displacement parameters by the data on jointing in terms of the Zapolyarny Mine. MIAB. Mining Inf. Anal. Bull. 2019, no. 10, pp. 106—116. [In Russ]. DOI: 10.25018/0236-1493-201910-0-106-116.

3. Sergunin M. P., Darbinyan T. P. Identification of rock mass jointing parameters in geological models in modern geoinformation systems (in terms of Micromine). Gornyi Zhurnal. 2020, no. 1, pp. 39—42. [In Russ]. DOI: 10.17580/gzh.2020.01.07.

4. Sergunin M. P., Darbinyan T. P., Shilenko S. Y., Grinchuk I. P. Digital surface modeling of an ore pass to reveal orientation of principal stresses and effect of rock fracturing. Gornyi Zhurnal. 2020, no. 6, pp. 28—32. [In Russ]. DOI: 10.17580/gzh.2020.06.04.

5. Ruehlicke B., Carter M. J., Ottesen C. G. The statistical eigenvector analysis technique (SEAT) for dip data analysis. Marine and Petroleum Geology. 2019, vol. 110, pp. 856—870. DOI: 10.1016/j.marpetgeo.2019.07.027.

6. Devis D. S. Statisticheskiy analiz dannykh v geologii [Statistics and data analysis in geology], Moscow, Nedra, 1990, 319 p.

7. Mardia K. Statisticheskiy analiz uglovykh nablyudeniy, per. s angl. [Statistics of directional data. English–Russian translation], Moscow, Nauka, 1978, 240 p.

8. Chini R. F. Statisticheskie metody v geologii, per. s angl. [Statistical methods in geology for field and lab decisions. English–Russian translation], Moscow, Mir, 1986, 189 p.

9. Woodcock N. H. Specification of fabric shapes using an eigenvalue method. Geological Society of America Bulletin. 1977, vol. 88, pp. 1231—1236. DOI: 10.1130/0016-7606(1977)88<1231:SOFSUA>2.0.CO;2.

10. Petri B., Almqvist B. S., Pistone M. 3D rock fabric analysis using micro-tomography: An introduction to the open-source TomoFab MATLAB code. Computers & Geosciences. 2020, vol. 138, article 104444. DOI: 10.1016/j.cageo.2020.104444.

11. Caumon G., Gray G., Antoine C., Titeux M. O. Three-dimensional implicit stratigraphic model building from remote sensing data on tetrahedral meshes: Theory and application to a regional model of La Popa Basin, NE Mexico. IEEE Transactions on Geoscience and Remote Sensing. 2013, vol. 3, no. 51, pp. 1613—1621. DOI: 10.1109/TGRS.2012.2207727.

12. Kokkalas S., Xypolias P., Koukouvelas I. K., Doutsos T. Relationships between Folding and Fracturing in Orogenic Belts: Examples from the Rhenohercynian Zone (Germany) and the External Hellenides (Greece). Geologica Carpathica. 2003, vol. 54, pp. 153—162.

13. Kirkwood С. Deep covariate-learning: Optimising information extraction from terrain texture for geostatistical modelling applications. ArXiv preprint. 2020, p. 14. DOI: 10.48550/ arXiv.2005.11194.

14. Mallik S., Bhowmik T., Mishra U., Paul N. Mapping and prediction of soil organic carbon by an advanced geostatistical technique using remote sensing and terrain data. Geocarto International. 2020, vol. 37, pp. 2198—2214.

15. Acharyya S. S., Mondal T. K. Magnetic shape fabric analysis from syntectonic granites: a study based on the eigenvalue method. Geological Magazine. 2022, vol. 160, рр. 222—234. DOI: 10.1017/S0016756822000747.

16. Rathmann N. M., Hvidberg C. S., Grinsted A., Lilien D. A., Dahl-Jensen D. Effect of an orientation-dependent non-linear grain fluidity on bulk directional enhancement factors. Journal of Glaciology. 2021, vol. 67, no. 263, pp. 569—575. DOI: 10.1017/jog.2020.117.

17. Whitaker A. E., Engelder T. Characterizing stress fields in the upper crust using joint orientation distributions. Journal of Structural Geology. 2005, vol. 27, pp. 1778—1787. DOI: 10.1016/J.JSG.2005.05.016.

18. Tavani S., Corradetti A., Matteis M. D., Iannace A., Mazzoli S., Castelluccio A., Spanos D., Parente M. Early-orogenic deformation in the Ionian zone of the Hellenides: Effects of slab retreat and arching on syn-orogenic stress evolution. Journal of Structural Geology. 2019, vol. 124, pp. 168—181. DOI: 10.1016/j.jsg.2019.04.012.

19. Elfwing Stefan, Uchibe Eiji, Doya Kenji. Sigmoid-weighted linear units for neural network function approximation in reinforcement learning. Neural Networks. 2018, vol. 107, pp. 3—11. DOI: 10.1016/j.neunet.2017.12.012.

20. Sosnin A. S., Suslova I. A. Neural network activation functions: sigmoid, linear, stepped, relu, than. Nauka. Informatizatsiya. Tekhnologii. Obrazovanie: Materialy XII Mezhdunarodnoy nauchno-prakticheskoy konferentsii [Scientific. Informatization. Technologies. Education: Proceedings of the XII International Scientific and Practical Conference], Ekaterinburg, RGPPU, 2019, pp. 237—246. [In Russ].

21. Tharwat A. Classification assessment methods. Applied Computing and Informatics. 2021, vol. 17, no. 1, pp. 168—192. DOI: 10.1016/j.aci.2018.08.003.