Application of hierarchical classification methods in a final class space formation in the task of determining rock types using well logging data

UDK: 550.832:681.518
DOI: 10.24887/0028-2448-2022-4-20-25
Key words: hierarchical clustering, solution tree, random forest, class space, machine learning, volumetric model, pre-Jurassic unit
Authors: V.Yu. Rudenko (MiMGO CJSC, RF, Moscow), D.E. Gurentsov (MiMGO CJSC, RF, Moscow), M.E. Smirnova (MiMGO CJSC, RF, Moscow), P.Yu. Kulikov (MiMGO CJSC, RF, Moscow), S.S. Gavrilov (MiMGO CJSC, RF, Moscow)

One of the basic units of the geological model of the field is a classification of rocks (lithotypes or petrotypes) determined by logging data. At the same time, in complex formations in case of a wide variety of classes, logging data not always can help to distinguish them by properties. In practice, some classes have fairly close physical properties despite differences in genesis, structure and texture. In this case, it is necessary to form an optimal set of classes for their further prediction and using in a geological modeling. Under this article an attempt of solving this problem for the acid Permian-Triassic igneous rocks of Western Siberia was made using a combination of random forest machine learning algorithm and hierarchical clustering. While hierarchical clustering, the basic matrix of features was built taking into account a volumetric model that was calculated using hybrid technology (simultaneous linear algebraic equations and genetic algorithm). As a result, the integration of technologies made it possible to solve the problem of reducing the class space for its further prediction using logging data. As a confirmation of the reliability of the applied approaches, a comprehensive analysis of the results was carried out. Thus, the results are approved by the petrographic analysis of the core data that shows the similar material composition and similar mechanism of formation of united classes. In the article conclusions about the possibility of using machine learning algorithms to solve similar problems are stated.

References

1. Rudenko V.Yu., Babakov I.V., Priezzhev I.I., Aplication of stochastic model and genetic algorithms for multimineral modeling based on principle of petrophysical inversion (In Russ.), Geofizika, 2020, no. 6, pp. 18–26.

2. Ward J.H., Hierarchical grouping to optimize an objective function, J. of the American Statistical Association, 1963, V. 58, pp. 236–244.

3. Sneath P.H.A., Sokal R.R., Numerical taxonomy: The principles and practices of numerical classification, San-Francisco: Freeman, 1973, 573 p.

4. Bar-Joseph Z., Gifford D.K., Jaakkola T.S., Fast optimal leaf ordering for hierarchical clustering, 2001, DOI: 10.1093/bioinformatics/17.suppl_1.s22

5. Breiman L., Friedman J., Olshen R., Stone C.J., Classification and regression trees, Wadsworth, Belmont, CA, 1984, 368 p.

6. Geurts P., Ernst D., Wehenkel L., Extremely randomized trees, Machine Learning, 2006, V. 63(1), pp. 3–42, DOI:10.1007/s10994-006-6226-1

7. Certificate of official registration of the computer program no. 2021610214. GSPetrophysics,   Authors: Rudenko V.Yu., Priezzhev I.I.

8. Rudenko V.J., Babakov I.V., Priezzhev I.I., Application of genetic algorithms for multimineral modeling based on the principle of petrophysical inversion, European Association of Geoscientists & Engineers - Conference Proceedings, Data Science in Oil & Gas, 2020, V. 2020, pp. 1–6, DOI: https://doi.org/10.3997/2214-4609.202054015

9. Rudenko V.J. Babakov I.V., Priezzhev I.I., Application of hybrid combination technology genetic algorithm using well-log data for multimineral modeling with computing of changes in each mineral endpoint, European Association of Geoscientists & Engineers – Conference Proceedings, Geomodel, 2020, V. 2020, pp. 1–5, DOI: https://doi.org/10.3997/2214-4609.202050083

10. Rudenko V.Yu., Gurentsov D.E., Gavrilov S.S., Smirnova M.E., Calculation of petrophysical inversion based on hybrid models in volcanic rocks of acidic com position of the pre-Jurassic complex of Western Siberia (In Russ.), Pribory i sistemy razvedochnoy geofiziki, 2021, no. 4, pp. 41–48.

 11. Ellanskiy M.M., Petrofizicheskie svyazi i kompleksnaya interpretatsiya dannykh promyslovoy geofiziki (Petrophysical relationships and complex interpretation of production geophysics data), Moscow: Nedra, 1978, 215 p.

12. Mitchell W.K., Nelson R.J., A practical approach to statistical log analysis, SPWLA 29thAnnual Logging Symposium, 1988, June 5–8.

13. Petrograficheskiy kodeks Rossii. Magmaticheskie, metamorficheskie, metasomaticheskie, impaktnye obrazovaniya (Petrographic Code of Russia. Igneous, metamorphic, metasomatic, impact formations), edited by Bogatikov O.A., Petrov O.V., Morozov A.F., St. Petersburg, Publ. of VSEGEI, 2009, 200 p.

One of the basic units of the geological model of the field is a classification of rocks (lithotypes or petrotypes) determined by logging data. At the same time, in complex formations in case of a wide variety of classes, logging data not always can help to distinguish them by properties. In practice, some classes have fairly close physical properties despite differences in genesis, structure and texture. In this case, it is necessary to form an optimal set of classes for their further prediction and using in a geological modeling. Under this article an attempt of solving this problem for the acid Permian-Triassic igneous rocks of Western Siberia was made using a combination of random forest machine learning algorithm and hierarchical clustering. While hierarchical clustering, the basic matrix of features was built taking into account a volumetric model that was calculated using hybrid technology (simultaneous linear algebraic equations and genetic algorithm). As a result, the integration of technologies made it possible to solve the problem of reducing the class space for its further prediction using logging data. As a confirmation of the reliability of the applied approaches, a comprehensive analysis of the results was carried out. Thus, the results are approved by the petrographic analysis of the core data that shows the similar material composition and similar mechanism of formation of united classes. In the article conclusions about the possibility of using machine learning algorithms to solve similar problems are stated.

References

1. Rudenko V.Yu., Babakov I.V., Priezzhev I.I., Aplication of stochastic model and genetic algorithms for multimineral modeling based on principle of petrophysical inversion (In Russ.), Geofizika, 2020, no. 6, pp. 18–26.

2. Ward J.H., Hierarchical grouping to optimize an objective function, J. of the American Statistical Association, 1963, V. 58, pp. 236–244.

3. Sneath P.H.A., Sokal R.R., Numerical taxonomy: The principles and practices of numerical classification, San-Francisco: Freeman, 1973, 573 p.

4. Bar-Joseph Z., Gifford D.K., Jaakkola T.S., Fast optimal leaf ordering for hierarchical clustering, 2001, DOI: 10.1093/bioinformatics/17.suppl_1.s22

5. Breiman L., Friedman J., Olshen R., Stone C.J., Classification and regression trees, Wadsworth, Belmont, CA, 1984, 368 p.

6. Geurts P., Ernst D., Wehenkel L., Extremely randomized trees, Machine Learning, 2006, V. 63(1), pp. 3–42, DOI:10.1007/s10994-006-6226-1

7. Certificate of official registration of the computer program no. 2021610214. GSPetrophysics,   Authors: Rudenko V.Yu., Priezzhev I.I.

8. Rudenko V.J., Babakov I.V., Priezzhev I.I., Application of genetic algorithms for multimineral modeling based on the principle of petrophysical inversion, European Association of Geoscientists & Engineers - Conference Proceedings, Data Science in Oil & Gas, 2020, V. 2020, pp. 1–6, DOI: https://doi.org/10.3997/2214-4609.202054015

9. Rudenko V.J. Babakov I.V., Priezzhev I.I., Application of hybrid combination technology genetic algorithm using well-log data for multimineral modeling with computing of changes in each mineral endpoint, European Association of Geoscientists & Engineers – Conference Proceedings, Geomodel, 2020, V. 2020, pp. 1–5, DOI: https://doi.org/10.3997/2214-4609.202050083

10. Rudenko V.Yu., Gurentsov D.E., Gavrilov S.S., Smirnova M.E., Calculation of petrophysical inversion based on hybrid models in volcanic rocks of acidic com position of the pre-Jurassic complex of Western Siberia (In Russ.), Pribory i sistemy razvedochnoy geofiziki, 2021, no. 4, pp. 41–48.

 11. Ellanskiy M.M., Petrofizicheskie svyazi i kompleksnaya interpretatsiya dannykh promyslovoy geofiziki (Petrophysical relationships and complex interpretation of production geophysics data), Moscow: Nedra, 1978, 215 p.

12. Mitchell W.K., Nelson R.J., A practical approach to statistical log analysis, SPWLA 29thAnnual Logging Symposium, 1988, June 5–8.

13. Petrograficheskiy kodeks Rossii. Magmaticheskie, metamorficheskie, metasomaticheskie, impaktnye obrazovaniya (Petrographic Code of Russia. Igneous, metamorphic, metasomatic, impact formations), edited by Bogatikov O.A., Petrov O.V., Morozov A.F., St. Petersburg, Publ. of VSEGEI, 2009, 200 p.



Attention!
To buy the complete text of article (a format - PDF) or to read the material which is in open access only the authorized visitors of the website can. .

Mobile applications

Read our magazine on mobile devices

Загрузить в Google play

Press Releases

29.06.2022
23.06.2022
17.06.2022