Application of Water Injection Profile Recognition Based on Machine Learning Method in F Oilfield

Abstract

Interlayer water injection profile is an important data for oilfield development and adjustment. At present, it is mainly based on field test, with high cost and few data. For the problem of water injection profile identification in F oilfield, this paper uses the dynamic and static data of reservoir to carry out the prediction of water injection profile by machine learning method. Based on the analysis of the influencing factors of water absorption profile, the sensitive parameters of 11 dimensions and their calculation methods are proposed, and the basic data sample database is constructed. XGBoost ensemble learning algorithm is used to realize small sample database prediction of longitudinal water injection profile in F oilfield. Compared with KH split water absorption method, it can more accurately grasp the change law of each water injection well's longitudinal water absorption status in the well area. The test results show that the recognition accuracy of the new identification method is 85.5% compared with the logging interpretation results. The water absorption index of each layer is predicted, and the water absorption difference between layers is clearly grasped, which points out the direction for injection production allocation.