From the date of the base of the Mining University passed 244 years 260 days.

Mining University master’s student designed the best machine learning algorithm for “Gasprom neft”

Thursday, 28 December, 2017
Mining University master’s student designed the best machine learning algorithm for “Gasprom neft”
Mining University master’s student designed the best machine learning algorithm for “Gasprom neft”

Vladislav Legkokonets won in the final stage of the contest for the design of the best machine learning algorithm for handling of applied modules of petroleum engineering. The contest was held by Moscow Institute of Physics and Technology and “Gasprom neft STC”. In addition to Russian participants ,the top graduates of Heriot-Watt University ( Edinburg, Scotland) took part in the contest.

Experts offered the finalists to solve t one out of six topical production problems by means of machine learning. Such criteria as precision of prediction, sequence and logicality in use of data processing and analysis algorithms as well as the possibility to apply the designed  model to more data were used for selecting the winner.

According to the judges at the contest the contestant who coped with the task best of all was master’s student of the chair of field development and operation of Mining University Vladislav Legkokonets (scientific supervisor Dmitriy Mordashev). Mars Hasanov, Director –General of OOO “Gasprom neft STC”, granted the prizes to Vladislav.

“Problems that were set to the participants of the contest were difficult for solving by means of classical approach, commented Vladislav on his achievement. “ I chose the problem in which it was necessary to reestablish velocity profile of shear waves along bore axis. Profile record by means of geophysical well logging is cost intensive. That is why it is preferable to design a tool for its reestablishment from basic logging suite. Machine learning permits to do it. Over half of finalists solved this very problem but the best was proved to be my solution.”

It is a reminder that in the process of application machine learning algorithms are good at solving not only a particular task but a lot of analogous ones.”Gasprom neft” intensifies efforts at implementation of this technology. Its introduction will permit to process enormous pools of operational data as well as to cut by half time for their analysis, minimize mistakes and influence of human factor. Besides company specialists will have a possibility to take into account new conformities found while developing  oil and gas fields , to select best practices for enhancement of oil recovery and to reduce costs.

Система Orphus
Стандартная версия сайта