Application of Fracture Unsupervised Pattern Recognition Technology in Y Gas Field

Accurate characterisation of fracture detail is fundamental to the fine description of a reservoir's structure and the deeper understanding of the reservoir. As oil and gas reservoirs gradually enter the middle and late stages of development, there is an increasing demand for fine identification of small fractures. However, due to the influence of burial depth and surface conditions, the current seismic data is more difficult to identify small fractures in the middle and late stages of development, and the identification accuracy is low. Combined with current methods of artificial intelligence big data analysis, this paper proposes an unsupervised mode fracture identification technique under superiority frequency conditions. The frequency that can reflect different scales of fractures is preferentially selected, on the basis of which a variety of different types of geometric fracture attributes are extracted, and then unsupervised pattern recognition algorithms are applied to allow the computer to automatically set and classify certain fractures with common characteristics by learning to compare, and to portray the spreading characteristics of single-scale and full-scale fractures, so as to improve the lateral discrimination ability of fractures and effectively enhance fracture recognition Through the application in the Y area, it has achieved fine fracture mapping within the gas reservoir, saving the interpreters' effort and time in analysing data, and obtaining ideal fracture pattern results, deepening the understanding of oil and gas reservoirs, and effectively supporting the evaluation of oil and gas reservoir development potential and later well deployment.