Prediction of Natural Gas Final Consumption using Artificial Neural Networks

Abstract

Natural gas is a kind of fossil fuel and itself is more suitable for reducing environmental pollution than the refined fuel oils such as gasoline(petrol) and diesel. Natural gas consumption is eco-friendly and is useful to fulfill energy demand for industries, transportation and other purposes. Myanmar is natural gas main producer in Asia and the majority of natural gas is exported to Thailand and China. In the near future, it is important to know the total consumption of natural gas for a country to fulfill the demand of country's needs. In this paper, Myanmar's natural gas final consumption will be predicted using Artificial Neural Networks(ANN). To predict natural gas final consumption for the coming years, actual recorded consumption data of Myanmar 1990–2015 are used. The last five years' data (2011–2015) are applied for testing and the previous years' data (1990–2010) are used for training. Population, gross domestic product(GDP), and other factors that affect natural gas final consumption are used as input for model building. The training model is strong with a minimum error rate of 0.005 Mean Squared Error (MSE). The developed ANN model is applied for the prediction of future natural gas consumption of Myanmar year by year. The proposed method is effective to predict Myanmar's natural gas final consumption and is useful in the studies of energy policy and ecological quality.