Day-ahead wind speed prediction by a Neural Network-based model
Accurate wind forecasting is valuable for a number of stake holders including farm, system and microgrid operators. The variability and non-linearity of the wind speed/power signal, compounded with the scarcity of time series data, constitute a challenge and make imperious the need of accurate and robust methods for wind forecasting. This paper presents a multi-variable model for day-ahead hourly wind speed/power prediction. The model is a combination of an input selection technique and a Neural Network (NN). First, the input selection technique selects the best set of inputs. Then, by means of the selected features, a NN forecasts the next values of the wind signal. The whole proposed method is examined on wind speed prediction of two wind farms to show the validity and accuracy of the proposed model. © 2014 IEEE.
Daraeepour, A; Echeverri, DP
2014 Ieee Pes Innovative Smart Grid Technologies Conference, Isgt 2014
Digital Object Identifier (DOI)