A hybrid approach to multi-step, short-term wind speed forecasting using correlated features

Wind power is becoming a main alternative energy source to meet the growing electricity needs. Forecasting wind speed important mitigate generation uncertainty and optimize asset utilization. This paper proposes hybrid prediction model with multivariate input multi-step output capability. The synthesizes linear time series regression nonlinear machine learning algorithm. neurons of are determined by number lag observations in autoregressive integrated moving average (ARIMA), also correlated meteorological features, such as direction, air pressure, humidity, dew point, temperature. further derived based on forecasting horizon. trained, validated, tested using 1.73 million hourly records from three cities diverse profiles. performance compared several existing methods root mean square error absolute error. Though does not show obvious advantage 1-h ahead prediction, it outperforms persistence model, ARIMA, univariate neural network models 3-to-24 h prediction. able reduce 20% comparison networks.