Entwicklung eines Rechenmodells zur Vorhersage der Windleistung fuer mittlere und grosse Versorgungsgebiete Abschlussbericht

In co-operation with partners out of industry and research a machine program was developed predicting the output of wind power plants. Three attributes should be realised by this prediction tool: Short computing time, usable for various grid regions, high reliability. Therewith the transmission system operators get a tool for reducing the amount of control energy which is needed to ensure the balance between power generation and consumption in their networks. This prediction tool for up to two days was developed exemplary for the northern grid area of the transmission system operator 'E.ON Netz GmbH' (ENE). The wind power prediction is based on numerical weather forecast from the German weather service (Deutscher Wetterdienst). The weather forecast is given for 16 representative sites within the ENE-area. The meso-scale model KLIMM (Klima Model Mainz) was used to calculate the meteorological variables near to the wind farms, which are connected to the one transformer substation belonging to one representative place. Therefor KLIMM is fed with the weather forecast given for one limited location in the representative sites. The transformation of the meteorological variables to the output of wind power plants at the representative site is done by Neural Networks. These Neural Networks have been trained with corresponding measurements. Using an existing online-model the total wind power for the whole ENE-area will be calculated from the individual wind power of the representative sites. The Evaluation of the prediction- and measured data from 2001 shows comparing with reference-models, that the prediction-model evolved in the project lead to very good results. (orig.)SIGLEAvailable from TIB Hannover: F02B1608 / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekBundesministerium fuer Bildung und Forschung, Berlin (Germany)DEGerman