Журналы →  Gornyi Zhurnal →  2022 →  №1 →  Назад

Название Testing machine learning algorithms as a case-study of gravity effect on helium emission
DOI 10.17580/gzh.2022.01.11
Автор Sergunin M. P.
Информация об авторе

NorNickel’s Polar Division, Norilsk, Russia:

M. P. Sergunin, Head of the Department for Geotechnical Supervision of Mining, Center for Geodynamic Safety, SerguninMP@nornik.ru


The Earth’s interior continuously experiences natural deformations. The connection between these processes and the changes in the gravity field is evident. For example, the gravitation of the Moon can distort water surface of a geoid by 0.5 m on the average, and has a direct effect on the crustal deformations. The geodetic data from GNSS show seasonal distortions of ground surface at the amplitudes of a few centimeters. During experimentation, it was assumed that variations in the gravitational field affect for the first turn jointing, which means natural opening or closure of joints. For this reason, the simplest method to estimate opening or closure o f joints is the measurement of concentrations of light volatile emissions from the subsoil, such as hydrogen, helium, methane and radon. These emissions always vary in concentration subject to the flow capacity of fractures which can both close or open in due conditions. A test object was selected with respect to the highest helium concentrations and dispersion. The tests used high-precision mass-spectrometric helium flow detector ASM 102S by Alcatel Vacuum Technologies, featured with high sensitivity and capability of automated data acquisition. The correlation between the variations in the gravitational field and the change in the helium concentrations was found. This connection was proved during processing a huge bulk of data by various machine learning algorithms. All in all, 27 models were learnt. In the course of testing the learning methods and configurations of neural networks, the learnt models were exercised in implementation of predictions.

Ключевые слова Stress state, permeability, joints, opening, closure, helium emission, Earth’s gravitational field, neural network
Библиографический список

1. Shabarov A. N., Tarasov B. G. On Effect of Terrestrial Tide Waves on the Stress-Strain State of Rock Masses. Journal of Mining Science. 2004. Vol. 40, Iss. 1. pp. 57–66.
2. Panzhin A. A. Experimental research and visualization of contemporary geodynamic movements. Problemy nedropolzovaniya. 2020. No. 3. pp. 32–39.
3. Altamimi Z., Rebischung P., Métivier L., Collilieux X. ITRF2014: A new release of the International Terrestrial Reference Frame modeling nonlinear station motions. Journal of Geophysical Research: Solid Earth. 2016. Vol. 121, Iss. 8. pp. 6109–6131.
4. Plate Motion Calculator. UNAVCO, 2020. Available: https://www.unavco.org/software/geodeticutilities/plate-motion-calculator/plate-motion-calculator.html (accessed: 19.09.2021).
5. Sashurin A. D., Panzhin A. A. Current problems and objectives in geomechanics. GIAB. 2020. No. 3-1. pp. 188–198.
6. Abbott B. P., Abbott R., Abbott T. D., Acernese F., Ackley K. et al. GW170814: A Three-Detector Observation of gravitational waves from a binary black hole coalescence. Physical Review Letters. 2017. Vol. 119, No. 14. 141101. DOI: 10.1103/PhysRevLett.119.141101
7. Nazarova L. A., Nazarov L. A., Skulkin A. A., Golikov N. A. Stress-permeability dependence in geomaterials from laboratory testing of cylindrical specimens with central hole. Journal of Mining Science. 2019. Vol. 55, Iss. 5. pp. 708–714.
8. Sergunin M. P. Neural network modeling of jointed rock mass behavior in topocentric terrestrial field: A case study of Oktyabrsky deposit. Gornyi Zhurnal. 2021. No. 1. pp. 69–74. DOI: 10.17580/gzh.2021.01.12
9. Raschka S., Mirjalili V. Python Machine Learning: Machine Learning and Deep Learning with Python, scikit-learn, and TensorFlow 2. 3rd ed. Packt, 2017. 770 p.
10. Aurélien G. Hands-on Machine Learning with Scikit-Learn, Keras & TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems. 2nd ed. Sebastopol : O’Reilly Media, Inc., 2019. 510 p.
11. Ross Z. E., Cochran E. S., Trugman D. T., Smith J. D. 3D fault architecture controls the dynamism of earthquake swarms. Science. 2020. Vol. 368, No. 6497. pp. 1357–1361.
12. 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. GIAB. 2019. No. 10. pp. 106–116.

Language of full-text русский
Полный текст статьи Получить