Method Of Analysising Carbonate Reservoir Image Log Facies And Its Application

As an important part of petroleum exploraton,carbonate reservoir has become the focus of research at present in China.Features in carbonate reservoir are complex.When we do research on semdimentary and structure of carbonate reservoir,besides the basic data such as core,the interpretation of imaging logging data is particularly important.Due to the high resolution imaging logging and long measurement,there were too much work when we did interpretation on the imaging logging data in a manual way.And the interpretation could be different on account of different interpreter.Along with the development of artificial intelligence technology,using computer for automatic interpretation of the imaging logging data became developing trend of logging interpretation in the future.In this paper,firstly we research and summarize the imaging logging facies of carbonate reservoir in the Tarim Basin,Ordos basin and Sichuan basin,establish a set of imaging logging facies identification charts,which is suitable for most carbonate reservoir in China.Then we use JAVA language on CIFLog to extract imaging logging features and facies models,and form a imaging logging facies model database for deep learning of computer.The convolutional neural network(CNN)is used for deep learning of imaging logging facies model data.Finally,it is applied to the carbonate reservoirs of changxing,maokou,qixia,longwangmiao and dengying formations in hechuan-tongnan area of Sichuan Basin to check if it works.By contrast verification,the accuracy reaches 84%.