Title :
Intelligent load forecasting techniques for local power suppliers
Author :
Bitzer, B. ; Roser, Frank
Author_Institution :
Divison Soest, Univ. of Paderborn, Soest, Germany
fDate :
Aug. 31 1999-Sept. 3 1999
Abstract :
This paper presents the results for daily load forecasts of a local power supplier. The new approach of this paper is to use three neural networks, each representing a period of a day, for the daily load forecast. Three commercial tools for neural networks are used to reduce the time for the development and tests. The neural networks are tested with real load values of a the power supplier for a period of 5 months including special days like public holidays, Christmas and New Years Day. A mean absolute percentage error less then 3% could be proved for the examined months.
Keywords :
electricity supply industry; load forecasting; neural nets; intelligent load forecasting technique; local power supplier; neural network; Data models; Forecasting; Load forecasting; Load modeling; Neural networks; Predictive models; Training; load forecasting; neural networks; rapid prototyping;
Conference_Titel :
Control Conference (ECC), 1999 European
Conference_Location :
Karlsruhe
Print_ISBN :
978-3-9524173-5-5
