Machine Learning and Geometric Mathematical Models in Kimberlite Well Classification Problems

Nowadays, the application of mathematical models in geology becomes more and more relevant. The steady trend towards the global digitalization has led to the possibility of using the most modern computational methods in the construction of mathematical models. Digitalization, further processing of digital data, their analysis and subsequent modeling contributes to the improvement of production efficiency. The purpose of this paper is the development of various methods of classification of kimberlite wells. The paper presents neural network, statistical and geometric mathematical models for solving the problem of kimberlite well classification. The problem was solved using geological and exploration data from wells drilled in the Süldükar and Ulakhan-Kurung-Yuryakh areas located in Western Yakutia. For the constructed models the estimations of the models' qualities were obtained, the comparative analysis of the models was carried out. The analysis of mathematical models showed that the most accurate models are neural network models and models using geometric methods. © 2023 ASSA.

Авторы
Gabdrakhmanova N. , Klimtsev P.
Издательство
International Institute for General Systems Studies
Номер выпуска
4
Язык
Английский
Страницы
179-187
Статус
Опубликовано
Том
23
Год
2023
Организации
  • 1 Peoples Friendship University of Russia (RUDN University), Moscow, Russian Federation
Ключевые слова
classification; curvature; five factor analysis; kimberlite wells; neural network; persistent diagram; topological data analysis
Цитировать
Поделиться

Другие записи