ANN based short term load forecaster

Abstract

Application of feedforward artificial neural network for short term load forecast is discussed in this paper. A single feedforward neural network architecture for a day ahead and week ahead forecast is addressed. The said architecture uses back propagation algorithm with momentum and adaptive learning for training. The performance of the network is tested on the UPSEB load data. The network performance for a day ahead and week ahead forecast is also compared. The marginal accuracy of the daily forecasts over weekly one is offset by single time training compared to seven times in case of a day ahead forecasts.