The Support Vector Machine (SVM) today is probably the most popular algorithm of machine learning, because of its ability to solve the tasks of regression, classification, image recognition, etc. A classification task involves categorizing data of researching object into certain class, and as large ranges of data are usually used, this object is represented in a n-dimensional categorical-continuous hyperspace. Increasing the number of space dimensions increases the accuracy of the classification, although it increases interpretation efforts of the obtained modeling results. In this framework, it is proposed to dramatically expand the range of features involved in petrophysical modeling, including such rarely used characteristics as chemical composition, secondary mineralogy, structural and texture parameters, etc. At the same time, the involvement of machine learning algorithms will increase the ability to predict the key petrophysical properties in oil and gas searching and exploration by obtaining classification rules with low errors and information losses. In the classification mode the Support Vector Machine indicates the expediency of using the porosity and the content of debris and cement to identify geological objects in which the supcapillary pores predominate in the pore space. The geochemical and the petrophysical features (resistance, velocity, porosity) best classify objects into a group with a parity representation of the pores in size. Low-capillary formations are poorly identified by the means of the Support Vector Machine.
Al-anazi, Ammal & Gates, Ian. (2010). Support-Vector Regression for Permeability Prediction in a Heterogeneous Reservoir: A Comparative Study. SPE Reservoir Evaluation & Engineering – SPE RESERV EVAL ENG. 13. 485-495. https://doi.org/10.2118/126339-PA.
Al-Anazi A.F., Gates I.D. (2012) Support vector regression to predict porosity and permeability: Effect of sample size, Computers & Geosciences, Volume 39, 2012, Pages 64-76, ISSN 0098-3004, https://doi.org/10.1016/j.cageo.2011.06.011.
Chollet, F., Kalinowski, T., Allaire, J. J. (2022). Deep Learning with R, Second Edition. Simon and Schuster.
Karpenko, O., Bulmasov, O. (2005). The application of neural networks technologies in the interpretation of geophysical research of wells. Geoinformatics. 1, 71–79.
Karpenko, O., Kucher, S., Ellanskyi, O. (2006). Expert assessment of optimality of the physical and geological model of rock based on the use of artificial neural networks. Prospecting and Development of Oil and Gas Fields. 3(20), 54-59.
Shabatura O. (2011). The using of algorithm of pattern recognition for petrogenetical systematization and classification of rocks. 10th International Conference on Geoinformatics: Theoretical and Applied Aspects. https://www.scopus.com/inward/record.uri?eid=2-s2.0-84907419491&partnerID=40&md5=30ea2e2a50ce1dbb4cb9577ffb5c3d03
Vyzhva, S., Onyshchuk, V., Onyshchuk, I., Reva, M., Shabatura, O. (2021). Porosity and permeability properties of consolidated rocks of the northern near edge zone of the Dnieper-Donetsk depression. Visnyk of Taras Shevchenko National University of Kyiv. Geology. 3(94), 37-45. http://doi.org/10.17721/1728-2713.94.04
Vyzhva, S., Onyshchuk, V., Onyshchuk, I., Reva, M., Shabatura, O. (2021). Petrophysical properties of the Lower Permian limestones of the Glinsko-Solokhivsky gas- and oil-bearing region of the Dnieper-Donets depression (DDd). 15th International Conference Monitoring of Geological Processes and Ecological Condition of the Environment. http://doi.org/10.3997/2214-4609.20215K2066
Vyzhva, S., Bezrodna, I., & Tkachenko, O. (2019). The structure of void space of complex structured Bashkir carbonate reservoir rocks from the Opishnianske deposit of the DDD. Geofizicheskiy Zhurnal, 41(3), 189–202. https://doi.org/10.24028/gzh.0203-3100.v41i3.2019.172472