Reservoir Permeability Prediction Based on Analogy and Machine Learning Methods: Field Cases in DLG Block of Jing’an Oilfield, China

Abstract Reservoir permeability, generally determined by experimental or well testing methods, is an essential parameter in the oil and gas field development. In this paper, we present a novel analogy machine learning method to predict reservoir permeability. Firstly, core test production data of other 24 blocks (analog blocks) are counted according DLG block (target block) Jing’an Oilfield, permeability parameters including porosity, shale content, thickness, saturation, liquid production, pressure difference optimized Pearson principal component analysis. Then, fuzzy matter element used calculate similarity between target analog blocks. According calculation results, predicted engineering (the relationship porosity QK-D7 similar (random forest, gradient boosting decision tree, light machine, categorical boosting). By comparing prediction accuracy two methods through evaluation index determination coefficient (R2) root mean square error (RMSE), CatBoost model has higher predicting with R2 0.951 RMSE 0.139. Finally, selected 121 wells block. This work uses simple logging quickly accurately without coring testing. At same time, results applied formulation development technology strategy, which provides new idea for application oilfield parameters.