Short-term wind power forecasting model based on Kalman filters and neural networks

  1. Ramirez-Rosado, I.J. 1
  2. Fernandez-Jimenez, L.A. 2
  3. Mendoza-Villena, M. 2
  4. García-Garrido, E. 2
  5. Lara-Santillan, P.M. 2
  6. Zorzano-Alba, E. 2
  7. Zorzano-Santamaria, P.J. 2
  1. 1 Universidad de Zaragoza
    info

    Universidad de Zaragoza

    Zaragoza, España

    ROR https://ror.org/012a91z28

  2. 2 Universidad de La Rioja
    info

    Universidad de La Rioja

    Logroño, España

    ROR https://ror.org/0553yr311

Libro:
Proceedings of the IASTED International Conference on Modelling, Identification, and Control, MIC

ISBN: 978-0-88986-782-6

Año de publicación: 2009

Páginas: 7-12

Tipo: Capítulo de Libro

SCOPUS: 2-s2.0-74549144239 GOOGLE SCHOLAR

Resumen

The important penetration of wind energy into the electric networks over the last years has propelled the development of wind power forecasting tools. The intermittent behaviour of wind and the necessity of supplying the electric demand require knowing in advance the electric power generated by wind farms. This paper describes a new wind power forecasting model, with forecasting horizons of six hours, based on the use of an original application of Kalman filters and suitable artificial neural networks. The measurement variables, needed by the Kalman filter, are replaced by estimated values from another related variable, allowing the use of the proposed forecasting model in wind farms where exclusively the electric power production is available online. The forecasting model is comprised of a Kalman filter and two genetically optimized neural networks. The proposed model has been applied to a real wind farm obtaining better results than those obtained from a set of representative forecasting models.