R.F. Miftakhov, P.A. Avdeev, G.N. Gogonenkov, A.K. Bazanov, I.I. Efremov
Mapping of faults based on machine learning and neural networks
Key words: seismic interpretation; faults; automation; Artificial Intelligence; machine learning; deep neural network; technique; software system.
For citation: Miftakhov R.F., Avdeev P.A., Gogonenkov G.N., Bazanov A.K., Efremov I.I. Mapping of faults based on machine learning and neural networks. Geologiya nefti i gaza. 2021;(3):123–136. DOI: 10.31087/0016-7894-2021-3-123-136. In Russ.
The stage of faults construction is one of the most important in a seismic interpretation cycle. Moreover, a process of fault tracking is rather time consuming and requirе a lot of human resources. Today, there are many technological analytical solutions aimed at automating the process of tracking the fracture surfaces. However, most of them have a number of limitations resulting from the fact that full automation of procedures in the work under the complicated geological conditions of the study area, as well as having low-quality seismic material is impossible. Thus, the problem of optimization of this process is still relevant for the production cycle of oil and gas and service companies. The paper discusses the results of the new approach to implementation of the fault automated mapping process based on the use of Artificial Intelligence through machine learning and deep neural networks. New algorithms implemented in the Geoplat Seismic Interpretation software system allow eliminating subjectivity as much as possible and considerably reduce time for structural interpretation of faults under different geological conditions.