Research On Key Technologies Of Oil Reservoir Data Modeling

At present,most oil fields in China have entered the "double-high" stage with high water containing and high recovery.The remaining oil is highly dispersed and difficult to exploit.With the development of artificial intelligence,making use of big data technology to take advantages of the massive data accumulated over the years,excavate the potential value in the data,and turn the traditional business-driven model to data-driven has great significance.However,due to the complexity of oil data has brought serious challenges to the construction of oil data;at the same time,the limitations of oil developers in the field of data science has made it difficult to apply machine learning technology in the oil field.In response to these problems,the article has launched the following research:(1)Expounding the ecological construction method of oil reservoir data from three aspects such as business process,data quality inspection and feature engineering,puts forward the new feature engineering methods for reservoir data.(2)Giving the key technologies for reservoir data modeling in the two application of reservoir connectivity analysis and high water consumption layer identification,including data sample construction,machine learning model construction,model application and evaluation.(3)From the perspective of software engineering,developing the reservoir big data management software which based on SOA and J2EE's high-availability technology architecture to provide a set of intelligent analysis algorithm tools and rapid application platforms for reservoir developers.