Half Hourly Electricity Load Prediction using Echo State Network

Shivani Varshney, Toran Verma

Abstract: Prediction of time series is a task that cannot be efficiently done by using feed forward neural network. Recurrent neural network are the suitable neural networks for time series prediction task. Electricity load series is also a kind of chaotic time series. In this paper a echo state network is proposed to calculate the electricity load after half an hour. No other type of data is used for prediction except the previous electricity load values. The predicted values obtained from various simulation runs are very close to the actual value.

Keywords: Neural Network, Time Series prediction, Echo State Network, Electricity load forecasting