Аннотации:
© SGEM2019. This paper describes theoretical and practical issues of neural network application for the seismic facies analysis. The authors attention was paid to the identification method of seismic facies by self-organizing neural network map by using the correlation matrix of calculated seismic attributes (Instantaneous frequency, Envelope, Gradient magnitude, Instantaneous phase, Variance, Relative acoustic impedance, Sweetness, Chaos) and facies from the wells. The method allows to build a facies distribution map. The purpose of this work was to study the examples of applying seismic facies analysis for the deposits of the Bobrikian horizon of one of the Tatarstan Republic oilfields, study its facial distribution and to give a conclusion about the facies changes on the area and give a prediction of the best areas for drilling new wells. As a result, authors obtained classification map, which was used for geological zoning and studying the facies change in the study area. Six facies classes were identified; three of which were opened by wells and were identified as Channel facies, Underwater slope facies, Underwater delta plain facies, Lagoon facies and the remaining 3 were assigned to the Floodplain facies. On the basis of the allocated facies, a conclusion was made on the facial situation of the Tatarstan Republic oilfield, and a prediction was given about best areas – the most promising areas for drilling wells is Channel facies. Underwater slope facies and Floodplain facies are not interesting for development.
