Artificial neural network-based genetic algorithm to predict natural gas consumption

Abstract

The present study developed a hybrid model to predict the natural gas consumption in the city of Yasouj in Iran. In proposed model, an artificial neural network-based genetic algorithm (ANN-GA) integrates weather records (degree-day, relative humidity, wind speed, rainfall) and real gas consumption to predict daily gas consumption. The GAis used to optimize the neural network topology and its parameters. The results indicate that the ANN-GA model show good agreement with real data with an absolute deviation of 2.19 % and a high correlation coefficient (\(R\) = 0.998). The proposed ANN-GA model can consistently and accurately be used to predict daily natural gas consumption and is a powerful tool that can assist authorities in developing equitable and economic policies for natural gas distribution.