Short-term load forecasting is a crucial step for proper operation of a battery energy storage system. In this paper, an artificial neural network forecaster is used for hourly based forecasting of the distributed power generation and load consumption. This paper focusses on using mutual information for the selection of training data for the artificial neural network models of the forecaster. The proposed approach reduces the forecasting error, especially after transients in the input-output mapping. Simulations with real data sets are executed to verify the effectiveness of the method.
|Title of host publication||Proceedings of the 47th International Universities' Power Engineering Conference (UPEC 2012), 4-7 September 2012|
|Place of Publication||Piscataway|
|Publisher||Institute of Electrical and Electronics Engineers|
|Publication status||Published - 2012|