Development index prediction of the early polymer flooding using SVM

Abstract

Field production data are generally nonlinear, small samples and even fuzzy. In order to predict dynamic production index accurately and immediately for early polymer flooding in Bohai oilfield, multi-step prediction model using support vector machine was established based on the idea of phase space reconfiguration of single variable time series. The built model was applied to run multi-step prediction on dynamic index, e.g. water cut and oil recovery factor, and the results showed the model had a good adaptive capacity and multi-step prediction performance, and provided an effective method to predict dynamic index of early polymer flooding in Bohai oil field.