Application of recurrent neural network for short term load forecasting in electric power system

In recent years multilayered feedforward networks with backpropagation learning algorithm have been extensively applied to short term load forecasting in electric power systems with very good results. In this paper we investigate the feasibility of applying recurrent neural network (RNN) for short term load forecasting. Different network architectures from fully recurrent (complete connectivity) to no feedback paths (only feedforward paths) are modelled and their characteristics for short term load forecasting are compared.