Published on: 2026-07-07
Source: Peter the Great St. Petersburg Polytechnic University –
An important disclaimer is at the bottom of this article.
At the Gazpromneft-Polytech Scientific and Educational Center, defenses of comprehensive final qualification papers by students were held as part of the strategic initiative “Project as a Final Qualifying Work (VKR)”. Practice-oriented training and preparation of final qualification works commissioned by industrial partners is one of the key tasks in transforming engineering education, implemented with the support of the federal program “Priority-2030”. The customer for all the projects was the Scientific and Technical Center of Gazprom Neft.
The project “Forecasting geophysical attributes based on seismic survey and well research data” was carried out by students of the Higher School of Theoretical Mechanics and Mathematical Physics of the Physical-Mechanical Institute of SPbPU. For the oil and gas industry, the development of this forecasting tool means reducing uncertainty at the stage of geological exploration work. The ability to look into the space between wells before drilling the next point allows refining the seismic stratigraphy, identifying zones of abnormally low velocities (hydrocarbon reservoirs), and optimizing the drilling grid. The results of numerical experiments showed that the solution proposed by the Polytechnic students has a lower error rate, while convergence is accelerated by 2-4 times compared to the classical approach.
The Gazprom Neft Research and Technology Center provided students with real data from geophysical well studies, seismic surveys, and reference synthetic models. The thesis consultant was Vyacheslav Kim, a leading specialist at the Competence Center for the Development of Integrated Asset Modeling of the Gazprom Neft Research and Technology Center. Mentors from Polytechnic — Olga Tsvetkova, chief project engineer of the Gazpromneft-Polytech Research Center and senior lecturer at WSHTMiMF, and Dmitry Pashkovsky, programmer at the Gazpromneft-Polytech Research Center and assistant at WSHTMiMF — ensured internal academic expertise for the project. Also, Angelika Zhuravskaya, a senior lecturer at the Higher School of Computer Technologies and Information Systems of the Institute of Computer Science and Cybersecurity, became a consultant on information systems methodology.
The team consisted of representatives from fundamental mechanics and IT fields. Darya Ter-Mekaelyan was responsible for developing the mathematical core of the FWI inversion model, implementing the adjoint state method, formulating the inverse problem, and verification on synthetic data, while Leonid Kuzyakin focused on software implementation, creating efficient neural network training algorithms in Python (PyTorch), and developing the user utility. A working prototype of the software utility was presented at the project defense.
Students of the Physical-Mechanical Institute and the Institute of Computer Science and Cybersecurity presented three ambitious projects at once.
As part of the development of “Predicting the Movement of the Drill String Based on LWD and Geophysical Well Logging Data from Vertical Boreholes,” students are creating a neural network generative model for the comprehensive interpretation of geophysical well logging data from vertical boreholes and logging-while-drilling (LWD) data from horizontal boreholes. The key idea is the simultaneous refinement of the geophysical reservoir model as new information becomes available and the prediction of the optimal drill string trajectory within the oil-bearing reservoir.
This project is carried out by students of two engineering departments at SPbPU: Ivan Troshin (specializing in “Mathematical Modeling of Oil and Gas Production Processes”) and Roman Kishko (specializing in “Information Systems and Technologies”). The consultant is Maxim Simonov, head of the Artificial Intelligence Development Center at Gazprom Neft Research and Technology Center. Scientific and methodological support is provided by Galina Ayupova, assistant at the Higher School of Theoretical Mechanics and Mathematical Physics at FizMekh and leading specialist at the Gazpromneft-Polytech Research Center, and Angelika Zhuravskaya, senior lecturer at the Higher School of Computer Technologies and Information Systems at IKNK.
Students of PhysTech and ICSN Azaliya Gataulina and Alexey Gerasimov are working on a comprehensive project “Automatic interpretation of geophysical well survey data based on hybrid neural network architectures.” The development is based on a hybrid neural network architecture that combines LSTM recurrent blocks to account for the vertical variability of properties, ChebyNet graph convolutions for spatial propagation of information between wells, and an Attention mechanism to adaptively weigh the contributions of various scales and data sources. This approach allows for consideration of both the deep trends of logging curves and the mutual arrangement of wells, which is especially important in sparse drilling grids.
The work on behalf of the university is supervised by lead researcher of the Gazpromneft-Polytech Research and Education Center, associate professor of the Higher School of Computer Technologies and Information Systems IKNC Sergey Khlopin. Students are advised by Sergey Bazhukov, chief specialist of the Competence Center for Integrated Asset Modeling Development at Gazprom Neft Research and Technology Center, and Igor Gruzdev, engineer at the Gazpromneft-Polytech Research and Education Center.
The results of numerical experiments within the project showed that the proposed hybrid model outperforms classical oil processing technologies (kriging) by 3–5 times in terms of mean squared error (MSE), depending on the well density. When the number of wells increases from 20 to 80, the error decreases most sharply—from 0.009 to 0.0012, and at 200 wells, it reaches values around 0.0004.
The Research and Technology Center of Gazprom Neft emphasizes the high practical significance of the work. The implementation of the domestic development will reduce dependence on imported software in the field of GIS interpretation and construction of three-dimensional cubes of petrophysical properties, especially at the stages of geological exploration with a sparse well grid.
The third project is the development of an innovative PNN model (probabilistic neural network) that determines the pressure drop gradient in the pipeline. Developing our own PNN model will reduce dependence on imported software and ensure reduced computational costs while increasing calculation accuracy.
An interdisciplinary team of students is working on the creation of a technological product under the guidance of experienced mentors — representatives from academia and industry: Ilya Sokolov (program “Mechanics and Mathematical Modeling,” profile “Mathematical Modeling of Oil and Gas Production Processes”) and Danil Donskoy (program “Systems Analysis and Management”).
Maxim Simonov, head of the Artificial Intelligence Development Center at Gazprom Neft Research and Technology Center, acted as a consultant for the practical part of the final qualifying work.
The “Project as a Thesis” format strengthens the position of the Gazpromneft-Polytech Research and Education Center as an engineering and personnel partner of the Gazprom Neft Research and Technology Center, engaging students with fundamental knowledge in the oil and gas industry and competencies in artificial intelligence.
Please note; This information is raw content obtained directly from the source of information. It represents an accurate account of what the source claims and does not necessarily reflect the position of MIL-OSI or its clients.