Application of a Multi-Scheme Ensemble Prediction System for wind power forecasting in Ireland and comparison with validation results from Denmark and Germany

摘要

A Multi-Scheme Ensemble Prediction System (MSEPS) for short-range forecasting of wind power has been applied and tested at various sites and areas. The MS-EPS is the first short-range ensemble prediction system that has been used for wind power forecasting. The ensemble technique is ideal for tackling the problems associated with wind power prediction error, because it provides physically meaningful information on the uncertainty of each forecast. This is a useful and necessary tool in the decision making process for electrical system operators or energy traders and/or markets. Forecasts with the MS-EPS system have been performed and analyzed at various locations over a four month winter period in 2005. Forecasts for a single wind farm in northwest Ireland indicate a Mean Absolute Error of installed capacity (nMAE) of 11.4%. Results from continental Europe were found to be 4.4% for Germany and 8.2% for Denmark when looking at the aggregated wind capacity. At the Horns Rev offshore wind farm, the nMAE was 14.5%. It is shown that the level of dispersion of wind power and the average load factor heavily influence the achievable accuracy of wind power prediction systems. The study also shows the wind power prediction error is reduced with a combination of increased wind farm dispersion and also increased number of wind farms. This result is an important finding for Ireland, where the electricity grid is operated with only weak interconnection to Northern Ireland and into Scotland, but where a high growth rate of wind power is expected in the coming years.