| Abstract [eng] |
Wind energy is one of the fastest developing renewable energy sectors in the world and in Lithuania. However, such rapid development of wind turbines is typically a challenge for the grid operator because it causes issues of grid balance and requires reserves. The solutions of this kind problems are wind power prediction systems and improvement of their accuracy. Therefore, the dissertation investigated the influence of different factors and carried out analysis of statistical and physical methods on wind power prediction, accurately estimated the influence of surface roughness and relief on wind speed variations and wind power forecasting accuracy, besides the most accurate predictive functions of wind power plants were determined. The paper also presents identification of the predictive power correction coefficients, as well optimal algorithms for short-term power forecasting, the most suitable parametric methods for mathematical wind power curves description were carried out. The analysis of the physical power forecasting methods revealed influence of wind shear and turbulence on the accuracy of power prediction processes. Considering to evaluation methods of long-term wind energy resources, the most accurate Weibull’s distribution parameters functions were determined at different wind speeds are as well. Considering to analysed factors and methods determining wind power prediction accuracy, new hybrid wind power forecasting method was developed, and it predicts wind turbine power by up to 4.7% more accurately, comparing to direct wind speed conversion from numerical weather prediction data. |
