ENERGY EDUCATION SCIENCE AND TECHNOLOGY PART A-ENERGY SCIENCE AND RESEARCH, cilt.30, sa.1, ss.15-20, 2012 (SCI İndekslerine Giren Dergi)
The basis for energy management is to estimate the demand for energy as accurately as possible and without error. There are many studies related to demand forecasting in the literature. In this study, the support vector machines and artificial neural networks models were used to estimate the natural gas demand of Turkey. The correct forecasting of natural gas consumption plays an important role for the amount of natural gas production, and to reveal import and export policies. A natural gas consumption model is developed by using the data from years between 1985 and 2000 where gross national product and population are independent variables in Turkey. This model is used to estimate the natural demand for years 2001 and 2006 and results are compared with real consumptions. Then, statistical analyses are conducted and the results are compared with studies in the literature. In conclusion, it is observed that support vector machines have less statistical error comparing to artificial neural networks for demand estimation of natural gas consumption in Turkey. The models which are obtained by using support vector machines are run for four different scenarios and the natural gas demand forecasts are obtained for Turkey until year of 2030.