Title :
Data mining techniques for very short term prediction of wind power
Author :
Vargas, Luis ; Paredes, Gonzalo ; Bustos, Gonzalo
Author_Institution :
Dept. of Electr. Eng., Univ. of Chile, Santiago, Chile
Abstract :
This paper presents a comparison of data mining techniques for wind power forecasting in a time frame out to 15 minutes ahead. The forecasting is focused on the power generated by the wind farms and the power changes are predicted by using multivariate time series models ARMA, focus time-delay neural network (FTDNN) and a phenomenological model of the turbines. All these models are tested with real data of a 18 MW wind farm.
Keywords :
autoregressive moving average processes; data mining; load forecasting; neural nets; power engineering computing; time series; wind power plants; wind turbines; ARMA; FTDNN; data mining techniques; focus time delay neural network; multivariate time series models; turbines; wind farms; wind power forecasting; wind power generation; Artificial neural networks; Forecasting; Mathematical model; Predictive models; Wind forecasting; Wind power generation; Wind turbines; Data Mining; Wind Power Forecasting;
Conference_Titel :
Bulk Power System Dynamics and Control (iREP) - VIII (iREP), 2010 iREP Symposium
Conference_Location :
Rio de Janeiro
Print_ISBN :
978-1-4244-7466-0
Electronic_ISBN :
978-1-4244-7465-3
DOI :
10.1109/IREP.2010.5563273
