404 Weather Central LP Wind Power Yield Forecast System

Monday, 7 January 2013
Exhibit Hall 3 (Austin Convention Center)
Justin Joseph Traiteur, Weather Central, LP, Madison, WI; and A. Rice, B. A. Wilt, S. J. Reinke, and R. Runnheim

Maximizing efficiency and conserving resources are key to the continued growth of the wind power industry. Weather Central LP's state-of-the-art Wind Power Yield Forecast System delivers an extremely accurate wind power yield forecast, maximizing operators' return on investment. The system utilizes Weather Central's proprietary Super MicroCast™ high-resolution numerical weather prediction model, customized for each individual wind turbine. The model incorporates each turbine's height, rotor specifications and exact location into the physical model. Unlike standard linear interpolation techniques, the Super MicroCast model forces the equations of mass, energy and momentum to the turbine's exact location for superior accuracy. In addition, Weather Central has developed an optimized neural network system that creates complex bias correction relationships between input variables and historical forecasts. Traditional neural networks have focused on 15-20 variables that have a direct impact on wind power forecasts. Because Weather Central owns the physical model, over five times as many meteorological variables can be derived and utilized as inputs into the neural network. This allows for a more accurate relationship between different atmospheric properties and an individual turbine's wind power production. The optimized neural network then computes which combinations of input variables are the best predictors of the future wind field. Therefore, the combination of the optimized neural network forecast with the Super MicroCast model allows for outstanding forecast accuracy. This presentation will present a detailed overview of Weather Central's Wind Power Yield Forecast System and accuracy metrics.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner