A novel falling model for wind speed probability distribution of wind farms

Abstract

Accurate description of wind speed probability distribution is an essential part of wind farm planning and dispatching. The two-parameter Weibull distribution is mostly used to fit unimodal wind speed probability distribution, but it suffers from low fitting accuracy and insufficient applicability. To solve these issues of the Weibull distribution, a novel distribution model, Falling Model, is proposed in this paper. In this model, the first Step is to “fall” the empirical wind speed distribution based on the Weibull. Then the probability distribution of falling form is fitted by a Cloud Model after segmentation. Finally, the segmented Cloud Models are superposed with Weibull to build the Falling Model. The goodness-of-fit (GoF) of the Falling Model, Weibull, Burr, and Gamma distributions was compared in a case study, and it showed that the Falling Model had the highest fitting accuracy. The sum of squared error of the GoF is 106 times better than the Weibull. Further, the applicability analysis of the Falling Model was performed using existing literature data, and it found that the Falling Model can be widely applied. The study shows that the unimodal distribution combines the advantages of both the Weibull and several Cloud Models.