Multi-Step Ahead Ultra-Short-Term Wind Power Forecasting Based on Time Series Analysis

Ultra-short-term wind power forecasting is helpful since it can mitigate the non-dispatchability and high intermittency of wind power generation, which can cause problems in the electric power system. This paper proposes a multi-step short-term wind power prediction model with step length of 15min based on the autoregressive time series model. WRF mode is combined with the AR model to simulate the wind speed in the future. The research is done for both steady wind and unsteady wind and their results are compared. The results indicate that AR model shows acceptable errors for steady wind prediction and it has better validity than that of unsteady wind. The results also reveal that the accuracy drops as the increase of prediction steps. For further study, multivariate fitting will be applied to obtain more accurate characteristic curves. Also, some variables related to wind power prediction will be considered in order to make better results.