Prediction of wind power generation based on Autoregressive Moving Average Model

In this paper, the time series of wind power generation from the Fujin Wind Farm located in China were used for this study. Autoregressive Model (AR) and Autoregressive Moving Average Model (ARMA) were set up by the Long Autoregressive method which is one of the time series analysis methods. In the process of modeling, three methods were used to determine model order. After analyzing these three different models, the weighted average algorithm was used to construct the ultimate wind power predicting model and the Normalized Mean Absolute Error (NMAE) is within 7%.