Research On Combined Forecast Of Wind Speed And Wind Power And The Uncertainty Analysis

Wind power is the most rapidly growing renewable power generation way both in China and worldwide.Due to the strong randomness and volatility of wind,large scale wind power integration will bring a great challenge to power grid stability and power quality,thus limit the development.Forecast of wind speed and wind power has become a hot interest in research.This paper studies in the ultra short-term forecast models and its uncertainty analysis based on the time series data of wind speed and power.The principles and models of three single forecast methods are proposed,including ARMA.BP-ANN and ANFIS model.Actual data from a wind farm is used to analyze the accuracy of these models and the effectiveness is proved.To improve the forecast accuracy,five combined-models are brought into the wind speed and power forecast.Numerical analysis shows a great effect of these combined-models and the superiority to single models.Main factors that bring out uncertainty in forecast process are introduced and classified.An uncertainty analysis model is established on the basis of quantile regression,which can be applied to analyze any short-term wind speed and power forecast models.Results prove the model effective and practical.To solve the disadvantages of single point forecast method put forward,an interval forecast method is established.The model is a combination of simple linear regression and GM(1,1),considering the complexity of related influence and lack of uncertainty factors data.A comparison between the linear regression model and combined model is given,validating the effectiveness of the model in wind power interval forecast.