Georgios Anastasiades

Since there is no efficient way to store wind energy, the wind power production decreases to zero if wind speed drops below a certain level. On the other hand, excessively strong winds can cause serious damage to the wind turbines, and hence they are automatically shut down at the 'disconnection speed', leading to an abrupt and quite unexpected drop of power generation. In addition, the wind power generated is limited by the capacity of each turbine. Therefore, it is quite important to produce accurate wind power forecasts for an efficient operation of wind turbines and reliable integration of wind power into the national grid. 

I am currently  working on wind power probabilistic forecasting using statistical and probabilistic  models. More specifically I am interested in the construction of various variability indices which are able to capture the volatile nature of a wind power time series, as a result of the different weather regimes. My research also focuses on quantifying the uncertainties related to the  wind power production under extreme weather conditions, which can cause very high wind speeds or wind lulls.

I am particularly interested in linear and nonlinear time series modeling, linear and nonlinear quantile regression,  Markov-Switching processes and extreme value theory.