Prediction models for wind energy production

During the resent years the number of wind mills in Denmark has increased dramatically. The amount of electricity produced by wind mills is now considerable compared to what is produced by other sources, and it is necessary to take this amount into consideration in the production planning. This is in particular the case for the areas Funen and Jutland where ELSAM is responsible for the production and delivery of electricity. Due to rather large startup periods for some production units it is advantageous to estimate the amount of wind energy up to about 36 hours ahead. In this project statistical methods for predicting the wind energy for the ELSAM area 36 hours ahead are developed. The methods are based on actual measurements from 7 wind mill parks distributed across the area. Methods for taking meteorological forecasts into the model are developed. Also models for on-line detection and correction of errors in the data are formulated. All the models are implemented with a graphical user interface, and the results will be used in the daily production planning at ELSAM. An off-line version of the program is developed for a distribution company called SEP in Holland.