Development of Realistic Demand side management strategies using artificial neural networks for the production of informative wind speed prediction signals

Abstract

Urgency to achieve large-scale integration of the stochastic wind energy production calls among others for the employment of novel solutions including energy storage, upgrade of electricity grids, and application of effective demand side management. To this end, the problem of limited wind energy integration becomes even more severe in isolated island regions, owed to the weak character of local electricity grids. For this purpose, the current study emphasizes on the need for the production of adequately reliable forecasting wind speed signals that can, in turn, inform the development of appropriate energy, and especially demand side management strategies. In this context, we use artificial neural networks (ANN) and provide prediction of wind speed for three different island locations of the Aegean Sea, evaluating the wind speed signals for 1-10 h ahead. Our results demonstrate that wind speed predictions up to even 3 h ahead can sufficiently inform the development of appropriate energy management strategies, designating the potential of ANNs in the field of wind speed prediction.