Short-Term Wind Speed Forecast Using Mathematical Morphology Decomposition and Support Vector Regression

This paper proposes a hybrid forecast algorithm to improve the accuracy of short-term wind speed forecast. Based on the nature of wind energy, a mathematical morphology decomposition method using erosion and dilation operators decomposition, is performed to decompose the wind speed data into two parts: mean trend component (MTC) and strong stochastic component (SSC). MTC has stable character and SSC is stochastic in a smaller time scale, which is the most important part that affects the forecast accuracy. Support vector regression (SVR) is adopted to make regression of MTC and SSC respectively. The proposed method is tested on a dataset of short-term wind speed to verify its validity. Two-day ahead forecasts are conducted and evaluated in four seasons. In addition, the correlation between window size and forecast accuracy is discussed. Simulation results are compared with persistence method and SVR method, which illustrate that the proposed model is of high prediction accuracy with a small amount of historic data.