Productivity evaluation method of offshore fractured reservoir based on artificial intelligence ChatGPT

Abstract At present, the analogy method is usually used to determine the production capacity. Due to the strong heterogeneity of fractured reservoirs in the Bohai Sea, the prediction error is large and cannot meet the needs of scheme deployment. Therefore, based on the test productivity, the characteristic parameter λ was introduced to characterize fractured reservoir productivity, and the relevant data affecting productivity were preprocessed by the K-nearest algorithm and Z-Score standardization method. Then, the feature vector of parameter λ was selected by the average impurity reduction MDI feature selection method. Finally, in the GPT Building module of ChatGPT, the reservoir parameters and productivity test data are used to train GPT and the deep forest prediction model with parameter λ is built intelligently by GPT. This method is applied to the CFD oilfield productivity evaluation in the Bohai Sea, and the prediction error caused by the difference in reservoir physical properties is reduced. The coincidence rate between prediction and actual is more than 85%, which provides a new idea and method for the productivity evaluation of offshore fractured reservoirs.