- Home
- Conferences
- Conference Proceedings
- Conferences
Data Science in Oil & Gas
- Conference date: October 19-20, 2020
- Location: Online
- Published: 19 October 2020
21 - 24 of 24 results
-
-
Ensemble probability model for short-term production forecasting
Authors V.S. Kotezhekov, K.I. Krechetov and D.K. KuchkildinSummaryCreating an ensemble of forecast models for short-term calculation of liquid / oil production in automated mode on a regular basis.
-
-
-
Dependence evaluation of machine prediction quality the structural surfaces by potential fields on the completeness of the training sample.
More LessSummaryThe work considers the methodology of machine prognosis the geometry basement surface, based on dependence regression analysis of the desired, fragmentary specified, parameter and known physical fields. The algorithm “Randon Forest”, implemented in the program “Tensor”, is used for the neural network prognosis. To compile the regression equation, are used transformants of potential fields (magnetic and gravitational) - gradients, regional, local components, statistical parameters calculated in sliding window. The prognosis of the structural basement surface was performed, which is set fragmentally along 2D seismic survey profiles. The observational error is assessed in the partial absence of data in the training sample. Complex application of seismic and non-seismic methods to solve structural problems in this approach provide a reliable prognosis, stable even with the lack of 50 percent of data training sample. According to the analysis results, the maximum error of the prognosis was 90 m, which corresponds to 5% of depth margin error. The algorithm prognosis is applicable to study the physical contrasting properties of structural surfaces at the low stage of territory exploration.
-
-
-
Analysis of parameters of oil and gas fields using Bayesian networks
SummaryIn this paper, the authors investigated the approach of Bayesian networks to the analysis of parameters of oil and gas fields. The study of existing approaches and methods of probabilistic modeling in relation to the problems of analyzing the parameters of oil and gas fields based on production data showed that the best approach should have a number of important properties: the interpretability of the model, the ability to work with various types of data, and the ability to process distributions of sufficient dimension in a reasonable time. Bayesian networks were chosen as the main tool of work, since they allow to develop models that are understandable to the specialist and allow to do this entirely on data with minimal involvement of expert knowledge. The experiments have shown that Bayesian networks are able to simulate multidimensional distributions of field parameters; the developed model can also be used to reconstruct data gaps and assess the significance of variables. A comparison was made of different architectures of Bayesian networks with different score functions. It also shows an example of assessing the information significance of the parameters of deposits.
-
-
-
Peculiarities of object model construction for automation systems for an oil and gas asset
Authors V.N. Ulyanov, D.N. Tokarev, I.I. Frolova, S.A. Frolov, N.K. Kayurov and K.S. SerdyukSummaryThe new approach for the creation of an ontological model of oil and gas producing automation, implemented in software. Modern oil and gas production enterprise is a multidimensional complex object presented as a set of processes (business and technological) affecting objects and their properties. Allocation of processes allows you to convey the key ontological characteristic - length and variability over time. This approach simplifies the process of development and expansion of a big software package by several teams, eliminating transaction costs.
-