ENBIS-17 in Naples
9 – 14 September 2017; Naples (Italy)
Abstract submission: 21 November 2016 – 10 May 2017
Empirical Bayes Strategy for Site-Specific Wind Forecasting from Short-Term Data
12 September 2017, 18:40 – 19:00
- Submitted by
- Antonio Lepore
- Antonio Lepore (University of Naples Federico II), Biagio Palumbo (University of Naples Federico II), Antonio Pievatolo (CNR-IMATI), Mauro Falcone (University of Naples Federico II)
- Developers and financiers of wind-farm projects highly demand for new methods that hasten the evaluation of site-specific wind potential for electrical energy production, which is known to be hampered by expensive long-term anemometric sampling. However, short-term wind data collected at candidate sites are usually poor when the wind is not blowing from the prevailing direction(s). The attention is focused not only on the speed but also on wind direction, because it determines the optimal wind-farm layout design, which aims to minimize the turbulent effects among adjacent wind turbines.
By means of a real case study, an empirical Bayes strategy is proposed and shown to be capable of enhancing the annual wind rose prediction at the given candidate site by integrating the short-term sample data with both historical information (from a neighbouring survey station) and expert opinion. Posterior wind direction and speed distributions are obtained marginally at first. In particular, the multinomial likelihood and its conjugate (Dirichlet) prior are adopted to obtain the posterior distribution of wind direction; whereas, the approach proposed in  is followed for the wind speed. Then, in this work, the posterior wind rose is drawn from the sample (with the given posterior marginals) obtained through non-parametric inverse transformation of the short-term (bivariate) data collected at the candidate site.
 Erto P., Lanzotti A., Lepore A. (2010), Wind speed parameter estimation from one-month sample via Bayesian approach. Quality and Reliability Engineering International, Vol. 26, Issue 8, pp. 853–862.
Return to programme