The Importance of Normalization in Predicting Dew Point Pressure by ANFIS

Abstract

The authors compared some transformation functions in order to normalize data points to predict dew point pressure for gas condensate reservoirs. The proposed models were developed using experimentally measured and collected data of 111 gas condensate samples covering a wide range of gas properties and reservoir temperatures. Best answer belonged to logarithmic transformation, which had an average relative deviation of −11.86% and average absolute deviation of 16.31% for data points used for test. The accuracy of the final model has been compared to prior models. These examples have shown that the model successfully captures the physical trend and is reliable.