Investigating using stochastic methods to generate training data for windpower prediction

This paper investigates the potential capability of stochastic methods to generate data for unndpower prediction purposes. Stochastic models have been used to develop data before, however, this paper shows that a simplistic single histogram model will not suffice for windpower purposes. The need to generate data is important as often there is only a short period of data available for use in developing a prediction system. This need is then exaggerated if information from multiple wind turbines is desired for the data input. © Institution of Engineers.