Using MLP neural network modeling to forecast daily natural gas consumption of Gorgan city
Volume Title: 1
Authors
Golestan Gas Company
Abstract
In this research, Gorgan city (Golestan province, Iran) was chosen as a case study and artificial neural network (ANN) modeling based on multilayer perceptron (MLP) networks was carried out to forecast the amount of daily gas consumption according to meteorological parameters. To this end, average daily temperature and relative humidity as well as season of the year were selected as inputs and optimum network architecture was determined through trial and error procedure. By investigating the network forecasting performance, network architecture with two hidden layers where the number of neurons was 18 and 20 for the first and second hidden layers, respectively, was chosen as the optimum architecture. In addition, “logsig” transfer function resulted in better network performance in comparison to “tansig” one. RMSE, MAE, and MBE errors for results obtained by using optimum architecture, were 1.306, 0.043, and 0.267, respectively.
Keywords
Subjects