Application of artificial neural network models and random forest algorithm for estimation of fracture intensity from petrophysical data

Abstract Natural fractures play an essential role in the characterization and modeling of hydrocarbon reservoirs. Modeling fractured reservoirs requires understanding fracture characteristics. Fractured zones can be detected by using seismic data, petrophysical logs, well tests, drilling mud loss history core description. In this study, feed-forward neural networks (FFNN), cascade feed forward (CFFN) random forests (RF) were used to determine density from logs. The model performance was assessed statistical measures including root mean squared error (RMSE), coefficient determination ( R 2 ), absolute (MAE), Kling Gupta efficiency (KGE) Willmott’s index (WI). Conventional good logs full-bore micro-resistivity imaging data available three drilled wells Mozduran reservoir, Khangiran gas field. According findings research, FFNN showed a higher KGE WI, correlation ) compared CFNN model. outperformed with lower neurons. models' also improved increasing number neurons hidden layers 8 35. study demonstrate that measured calculated intensity is excellent agreement image log results showing 92%. RF algorithm stability robustness predicting 93%. successfully as aid more successful reservoir dynamic production analysis.