A post-processing Kalman Filter applied on wind power forecasting at S?o Vicente Island

摘要

This work proposes a model for predicting wind power, which combines the data provided by a Numerical Weather Prediction model and an adaptive statistical technique Kalman Filter. A post-processing linear Kalman filter based on ARX modeling is used over data produced by a Numerical Weather Prediction model to further improve wind power forecasting. A wind farm located on the island of S?o Vicente in the Cape Verde Archipelago, is used as a case study to test the model. The proposed model was able to reduce by a half the wind power prediction errors.