Method of Forecast Wind Speed Based on Wavelet Analysis and Quantile Regression

Because of wind speed with the randomness and non-stationary properties, wind power has a great effect when it integrates to power grid. We can only prediction a more accurate wind speed to reduce its harmful of the power grid. However, it is difficult to make more accurately forecast because of its characteristics. This article we discuss a method that wind speed processed by wavelet at first, then giving a different quantile points forecast of each layer of wind speed decomposition using the quantile regression, finally we gain the differential forecast by reconstruction every layers predict. The result shows this method prediction effect is better than quantile regression forecast and ARMA model prediction, especially in the extreme points.