Predictive Model for Estimating Wind Farm Power Output

This work focuses on improving the accuracy of electric power production pre- dictions for wind power plants. These predictions can be used for dispatching generation and identifying under-performing turbines for maintenance. To predict the wind speed, historical data was sorted into regimes, and a hidden- Markov model was created to model wind speed switching between them. In steady state, the relationship between wind speed and electric power production can be described by a non-linear power curve, but this does not capture transient effects. This work proposes a model that incorporates a type of autoregressive function whose parameters are easier to estimate than a complex physical model. Data from a wind power plant was obtained from a non-utility generation company for creation and verification of the models.unpublishednot peer reviewedU of I OnlyUndergraduate senior thesis not recommended for open acces