A hybrid ARIMA-DENFIS method for wind speed forecasting

This paper proposes a hybrid autoregressive integrated moving average - dynamic evolving neural-fuzzy inference system (ARIMA-DENFIS) model for wind speed forecasting. The theory of ARIMA, DENFIS and the hybrid of the two are discussed. The proposed model is evaluated with NDBC wind speed data and the results show that the proposed hybrid ARIMA-DENFIS model outperforms DENFIS model in most of the cases. It has comparable or better error measures than ARIMA model. In addition, when the forecasting horizon increases, the advantage of the proposed ARIMA-DENFIS model becomes more significant.