Hourly Electric Load Forecasting Algorithm based on Echo State Neural Network

An algorithm for hourly electric load forecasting based on echo state neural networks (ESN) is proposed in this paper. ESN is a new paradigm for using recurrent neural networks (RNNs) with a simpler training method. While the prediction, load patterns are treated as time series signals; no further information is used than the past load data records, such as weather, seasonal variations. The relation between key parameter of the ESN and the predicting performance is discussed; ESN and feedforward neural network (FNN) are compared with the same task also. Simulation experiment results demonstrate that the proposed ESN algorithm is valid and can obtain more accurate predicting results than the FNN for the short-term load prediction problem.