Abstract:

Wind speed prediction is of great significance to wind power plant scheduling and control. In view of the randomness and intermittence of wind speed series, an EMD-GWO-SVR combined prediction model is proposed in this study. Empirical mode decomposition is performed on the original sequence, and grey wolf optimization algorithm is used to optimize the parameters of the support vector regression model. Then, the optimized parameters are substituted into the support vector regression model, and the decomposed eigenmode function and residual term are predicted, respectively. The predicted results are added together to predict the wind speed. Taking the historical meteorological data of Jiuquan in Gansu province as an example, six forecasting models including BP neural network, SVR, PSO-SVR, GWO-SVR, EMD-PSO-SVR and EMD-GWO-SVR, are established to forecast the wind speed. The simulation results show that the prediction accuracy of the proposed EMD-GWO-SVR model is 61.759 8% higher than that of SVR, and the evaluation values of MAE, MAPE and RMSE error indexes are significantly lower than those of the other five models.

Key words: wind speed forecasting, BP neural network, empirical mode decomposition, particle swarm optimization algorithm, GWO algorithm, parameter optimization, support vector regression, prediction accuracy