Quality Controlling Wind Power Data for Data Mining Applications

Wind power generation is playing an increasingly important role in worldwide energy production. In order to optimize the utilization of wind power, it is critical to have a good handle on observed winds, the associated power production at wind farms and the power delivered to associated connection nodes. In practice, the power production at various wind farms is subject to wind farm curtailment, high speed wind turbine tripping, production loss due to turbine icing, turbine availability, and so forth. As a result, power production at wind farms can deviate significantly from the industrial wind power curves supplied by the turbine manufacturers.

This presentation will cover various challenges in the quality control of wind power production data. It will then discuss inter-percentile range techniques that can be used to filter the production data so that the filtered data can subsequently be used in automated wind power forecasting algorithms. Finally, it will discuss the sensitivity of wind to power conversion based on the application of different levels of quality control to the underlying wind power production data.