Short-Term Wind Power Generation Forecasting Based on the SVM-GM Approach

Short-term wind power generation forecasting plays an important role in power generation as it prevents the power system operation from uncertain and intermittent nature. This article proposes a novel hybrid method comprising the support vector machine and gray model (SVM-GM). Firstly, SVM is used to forecast the wind speed and direction of four seasons in the year according to the different weights calculated by gray correlation analysis through the use of the weather data obtained from the historical data. Then, the wind power generation is predicted by gray model in terms of the combination of the predicted numerical weather and wind information predicted by SVM, the particle swarm optimization is introduced to continuously optimize the background value of gray model to achieve more accurate prediction curves. Finally, the Fourier residual sequence is included to provide a deterrent effect of uncertainties of environment. Thereafter, the proposed approach adopts the data collected in an actual wind farm. Simulation results with four seasons validate that the proposed approach performs more effectively than other eight classical prediction methods in forecasting the short-term wind power generation of the wind farm.