Combining Neural Networks and Evolutionary Algorithms on Energy Demand Prediction

Energy demand prediction is very important to adopt appropriate policies. The reliability of intelligence and non-linear methods، particularly neural networks and evolutionary algorithms on energy demand prediction، hasbeenproved in numerous studies. This is due to volatility and non-linearity of energy demand and the affecting variables. Despite their lot of strengths، these techniques are faced with important issues such as imposing specific forms in evolutionary algorithmor necessity of large samples in neural networks. The purpose of this study is to exhibit a combined algorithm for more accurate energy demand prediction to overcome the disadvantages of individual techniques and to use their benefits simultaneously. To this purpose،the performance of various techniques in energy demand prediction has been investigated during 1967-2011. The results indicate that the neural networks learned by evolutionary algorithms in terms data limitations have desired outcomes، and the neural network base on the combination of genetic and particle swarm algorithms combination provides very good results. Comparisonof the results of this study with those of other studies in this field indicates the higher ability ofthe proposed algorithm and further confirms the high explanatory power of the used variables. Moreover، future energy demand projection indicates that the energy demand is going to be 1817، 1643 and 1457 million barrelsof oil equivalent (BOE) in 2025 under three different scenarios.