•  
  •  
 

Keywords

wind power, prediction error analysis, mixed t Locationscale distribution, improved Kmeans clustering

Abstract

Characteristics analysis of wind power prediction error can provide more accurate reference for optimal dispatch and stable operation of power system. This paper proposes the mixed t Locationscale distribution model to describe the probability distribution of wind power prediction error characteristics quantitatively. Then it uses improved Kmeans clustering algorithm to determine the model parameters. The distribution characteristics of the ultrashortterm prediction errors of wind power under different prediction methods are validated and analyzed with the measured data of a wind farm. Based on the measured data of the wind farm, we predict and analyze the errors produced by the two prediction models of time series and support vector machines, respectively. It is verified that the model can effectively describe the probability distribution of prediction errors.

DOI

10.19781/j.issn.16739140.2020.05.015

First Page

111

Last Page

118

Download

Share

COinS