Reservoir characterization is a scientific and technical problem full of uncertainty,since oil and gas reservoirs are extremely complex underground systems.The traditional theories and methods of reservoir characterization,mainly driven by knowledge,have inherent limitations and encounter numerous development bottlenecks.In recent years,artificial intelligence,especially deep learning,has been widely applied in reservoir characterization research,and breakthroughs have been made in specific problems.However,research in various problems is relatively dispersed,lacking systematic theory,and the integration of artificial intelligence and domain knowledge is insufficient.This paper proposes the concept and method of intelligent reservoir characterization driven by hybrid knowledge and data.Based on the characteristics of reservoir data,advanced artificial intelligence technology is used to integrate domain knowledge,fully mining the useful information hidden in big data,in order to achieve more reliable,high-precision,and efficient reservoir characterization.This paper summarizes the latest research progress in the key research content of reservoir intelligent characterization,mainly including deep learning-based reservoir logging evaluation,seismic reservoir prediction,and stochastic reservoir modeling.Finally,the development prospects of intelligent reservoir characterization are proposed.
关键词
油气藏表征/油气藏大数据/知识驱动/数据驱动/人工智能/深度学习
Key words
Reservoir characterization/Reservoir big data/Knowledge-driven/Data-driven/Artificial intelligence/Deep learning
维普
万方数据