Open AccessMethodology and precision research of wind farm power prediction
Author(s) -
Rimantas Pranas Deksnys,
Aldas Stankevicius
Publication year - 2017
Publication title -
elektronika ir elektrotechnika
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.224
H-Index - 26
eISSN - 2029-5731
pISSN - 1392-1215
DOI - 10.5755/j01.eie.23.1.12837
Subject(s) - linear regression , regression analysis , wind power , wind speed , statistics , mathematics , range (aeronautics) , exponential function , variables , simple linear regression , regression , variable (mathematics) , power (physics) , transformation (genetics) , econometrics , meteorology , engineering , geography , mathematical analysis , physics , biochemistry , electrical engineering , chemistry , quantum mechanics , aerospace engineering , gene
The article presents a newly developed statistical regression wind farm power change prediction model. Results of the research and the data analysis performed show that the model is able to evaluate factors determining the wind farm gross output and to increase the prediction accuracy. The influence of the regression equation independent variables on the dependent one is determined by the means of the Students ttest, and the levels of varying equation coefficients are established. The transformation of weather density and wind speed product corresponding to the linear stochastic dependence of the variable power characteristic part for the wind power plant is estimated. Expression of the transformation is suitable to use for predicting the wind farm power in the range from the minimum values to the installed ones. The statistical regression model of the wind farm power prediction is presented on the basis of given technique of the linear regression analysis, the exponential regression equation, and variable coefficients of regression equation. Results of power prediction by the given model precision research show that the one-day relative average prediction error does not exceed 7.52 % of the installed value