Application of Data Mining Methods for Power Forecast of Wind Power Plants

Since the last decade power systems underlie a drastic change due to increased exploitation of renewable energy resources (RES) such as wind and photovoltaic power plants. A result of this process is a significant increase of fluctuating generation in low, middle and high voltage grids. Consequently, impacts on short and middle term capacity planning of power plants occur and must be handled to avoid imbalances between generation and demand at any time. Therefore, forecasts of wind and photovoltaic generation play a very important role. Quality improvements potentially ease planning and lead to cost reductions. This work investigated the dependencies of input parameters. The optimal parameter selection was achieved through application of data mining methods. Finally, the wind power prediction was demonstrated with Artificial Neural Networks and Physical Models.