In recent years, the wind speed forecasting becomes more and
more important in the safety of wind energy [1]. The wind speed
predictions can decrease the possibilities of wind power breakdown
and protect the security of the wind power integration [2].
However, since the wind speed signals are non-stationary so that
it is difficult to realize their high-precision predictions. Aimed at
the wind speed short-term forecasting, some scientist have presented
important forecasting strategies. Based on the time horizons
[3], the wind speed forecasting methods include very shortterm
ones (from few seconds to 30 min head), short-term ones (from
30 min to six hours ahead), medium-term ones (from six hours to one
day ahead) and long-term ones (from one day to one week or more
ahead). Based on the computational mechanism, the wind speed
forecasting methods can be classified into three types as: statistical
methods, physical methods, intelligent methods and hybrid methods