1 / 24

Dr. Ulrich Focken EWEC 2007 MILAN, 08. May 2007

>>>. Optimal Combination of different Wind Power Predictions. Dr. Ulrich Focken EWEC 2007 MILAN, 08. May 2007. Overview. Motivation Weather Models Classification and Combination Summary. Motivation.

fisk
Download Presentation

Dr. Ulrich Focken EWEC 2007 MILAN, 08. May 2007

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. >>> Optimal Combination of different Wind Power Predictions Dr. Ulrich Focken EWEC 2007 MILAN, 08. May 2007

  2. Overview • Motivation • Weather Models • Classification and Combination • Summary

  3. Motivation • Wind power prediction systems commonly use only one single numerical weather prediction model (NWP). • But everyday experience shows: NWP models have strengths and weaknesses in different situations. • Our approach: Optimal combination of weather models adapted to different weather situations.

  4. Combining Europe‘s NWP for a better forecast

  5. Domains overlap

  6. Previento – the physical approach Previento • Physical Model: • Spatial refinement • Thermal stratification • Regional upscaling • Forecast uncertainty

  7. rmse root mean square error (rmse) dayahead-forecast January-October 2005 for single forecasts

  8. use weather information as expert input Rule-based combination is required • “Combine and average: [...] Simple average performs as well as more sophisticated statistical approaches.” Clemen, R.T., Combining forecasts: A review and annotated bibliography, Int. Journal of Forecasting 5 (1989) 559-582. • „Rule-based forecasting: [...] We believe that this procedure will lead to improvements.“ Armstrong,J.S., Combining Forecasts: The End of the Beginning or the Beginning of the End?, Int. Journal of Forecasting 5 (1989) 585-588.

  9. Combination of wind power forecasts Model 2 Model 1 Model 3 Model n etc. ... + model X + Previento + Previento + Previento combination tool 1. classification of weather situation 2. optimal combination combined wind power prediction „CombiTool“

  10. How the CombiTool works 1. Classification – Find significant weather situations • Principal Component Analysis Simplifying the dataset of meteorological parameters by reducing multi-dimensional data set to lower dimensions • Clustering Clustering groups similar objects into different subsets (clusters), so that the data in each subset share some common trait. here: similar weather situtation 2. Optimal Combination – Find the best combined forecast Find in each situation (cluster) the optimal weighting factors

  11. low passing North high pressure Results Clustering:Mean of u- und v-component and pressure in clusters pmsl [mbar]

  12. Weather situation „Cyclone passing – type A“

  13. Weather situation „Cyclone passing – type A“

  14. Cyclone passing – type A : one model is delayed power [% inst. power] days

  15. High pressure Eastern Europe

  16. High pressure Eastern Europe: models differ power [% inst. power] days

  17. Combination 18. Juli 2005 (+1d,2d,3d)

  18. Optimal factors differ from situation to situation normalized average combination factors [%]

  19. Accuracy in individual weather situations • using optimal weights for each weather situation leads to considerable improvement 5.0 % 3.9 % overall rmse rmse [% inst. power] Cyclone passing – type A High pressure Eastern Europe best single model sitation based combination

  20. Combination in extreme events combination power [% inst. power] days • combination very benefitial in extreme events

  21. Summary • NWP have strengths and weaknesses in different weather situations. • Just putting together forecasts is not sufficient, careful selection needed. • Automatic classification scheme based on methods from synoptic climatology generates useful weather classes. • Optimal combination based on weighting factors for specific weather situations outperforms individual forecasts. • Combination avoids large forecast errors in extreme events

  22. Summary • The system will go in operation at in Juli • It will use at least 8 forecasts from 4 different forecast provider as input

  23. Thanks for your attention !

  24. >>> ContactDr.Ulrich Fockenenergy & meteo systems GmbHMarie-Curie Straße 126129 Oldenburgulrich.focken@energymeteo.dewww.energymeteo.de

More Related