Wind Speed Interval Prediction Based on Lorenz Disturbance Distribution

Abstract

Wind speed is one of the fundamental influences affecting wind power uncertainties, and these uncertainties influence the stability of wind power grid. Therefore, to quantify the uncertainty fluctuation risk of wind power and reduce the uncertainty in the process of correcting wind speed, the wind speed interval prediction based on Lorenz disturbance distribution is presented. First, the point prediction of wind speed is obtained by an RBF neural network and empirical mode decomposition. Then, Lorenz disturbance sequence (LDS) with nonlinear and strong fluctuation is introduced, and the probability density distribution of LDS is estimated by Bootstrap and kernel density method. Next, the probability density distribution of LDS and quantile regression is, respectively, adopted to carry out wind speed interval estimation. Finally, comparing the reliability indexes and sensitivity indexes of those methods, the validity and stability of the proposed method can be verified. The results show that wind speed interval prediction based on the distribution of LDS can describe the fluctuation rage and tendency of wind speed effectively, and its result is stabler than that of quantile regression. In this research, the major differences between this paper and recently published manuscripts lie in introducing the distribution of LDS to make wind speed interval prediction; LDS can quantify the uncertain factors in the atmosphere and to reduce the uncertainties of prediction process, predicted interval of wind speed can provide more scientific reference for wind turbine to make start–stop scheme, which helps to quantify uncertain risk of wind power and ensure the stable operation of power system.