Wind Speed Prediction Based on Improved Self Excitation Threshold Auto Regressive Model

As a kind of clean energy, wind energy has become the focus among the world because of its rich reserves and easiness of exploitation and utilization. In order to reduce the abandoned of wind resource, stabilize the power quality and realize the power grid dispatching, The power grid needs to know the changes of the wind speed at every moment and accurately predict the change of the wind speed at the next moment, so the high quality wind speed forecast is more and more important. The analysis shows that the wind speed series has obvious nonlinear characteristics, so the traditional linear model is not accurate enough. In this paper, an improved nonlinear self excitation threshold autoregressive (SETAR) model, which is the self excitation threshold autoregressive moving average (SETARMA) model, is proposed to predict the wind speed. A detailed modeling method of SETARMA model is presented, and the method of testing the nonlinearity of wind speed series is discussed in this paper. The model is verified by a case study of a wind power plant in Spain, the simulation results show that compared with the traditional ARMA prediction methods, the wind speed prediction based on SETARMA model is better and has a higher accuracy.