Deriving Wind Speed Distributions Based on Bayesian ModelAveraging Method

摘要

In this paper, Bayesian model averaging (BMA) method is, for the first time, applied to model wind speeddistributions for the purpose of estimating long term wind energy potentials. The derived BMA probabilitydistribution is an average of the distributions of the ensemble models weighted by their posterior probabilities giventhe data of observations. The data used in this study are the mean hourly wind speeds over two years collected atmultiple sites in North Dakota. While none of the statistical models tested is universally plausible for all the sites,the BMA model demonstrates its robustness, reliability and comparative accuracy in estimating the wind speeddistributions for all sites. A simulation study is also conducted and the result shows that the BMA model has smallergoodness-of-fit values than each single model, indicating that the BMA model is comparatively accurate andreliable.