Development of methodology for evaluting forecasting models in aeolic energy

This thesis deals with the development of a methodology for evaluating forecasting models in wind energy. In particular, the effect of the structure of the wind turbines’ power curve on the efficiency of the wind speed forecast models is examined, especially in terms of the provided energy reliability and the economic efficiency of the respective wind system. In this context, a simulation of wind energy production through a standard wind turbine with input data of historical hourly wind speed time series, as well as corresponding time series of hourly wind speed forecasts, is constructed. A thorough statistical analysis of the wind speed time series data is performed, while the efficiency of wind forecasts is compared with the corresponding energy efficiency, through the application of a variety of statistical performance measures. Subsequently an empirical pseudo economic index that derives from a simple yet effective pricing procedure of the energy deficits and surpluses is proposed. Thus, conclusions are drawn regarding the forecast’s reliability based on the critical parameters of the achievement of energy demand and the economic viability of the wind energy production process. Finally, utilization of the methodology is proposed, both in problems of wind forecast and in matters of optimization of the design and operational decisions of wind systems.