Estimation of the non records logs from existing logs using artificial neural networks

Finding the information of the hydrocarbon reservoirs from well logs is one of the main objectives of the engineers. But, missing the log records (due to many reasons such as broken instruments, unsuitable borehole and etc.) is a major challenge to achieve it. Prediction of the density and resistivity logs (Rt, DT and LLS) from the conventional wire-line logs in one of the Iranian southwest oil fields is the main purpose of this study. Multilayer neural network was applied to develop an intelligent predictive model for prediction of the logs. A total of 3000 data sets from 3 wells (A, B and C) of the studied field were used. Among them, the data of A, B and C wells were used to constructing and testing the model, respectively. To evaluate the performance of the model, the mean square error (MSE) and correlation coefficient (R2) in the test data were calculated. A comparison between the MSE of the proposed model and recently intelligent models shows that the proposed model is more accurate than others. Acceptable accuracy and using conventional well logging data are the highlight advantages of the proposed intelligent model.