Short-term wind speed prediction model of VMD-FOAGRNN based on Lorenz disturbance

Abstract

Accurate wind speed prediction is the key to wind energy development and utilization. However, due to the intermittent, random and chaotic nature of wind speed, wind speed prediction has the problem of low accuracy. So a short-term wind speed prediction model of VMD-FOAGRNN based on Lorenz perturbation is proposed in this paper to predict wind speed. Firstly, the actual wind speed sequence is decomposed into multiple components by variational mode decomposition (VMD). After that, the generalized regression neural network (GRNN) model is used for training and prediction for each component, and the GRNN model is optimized by the fruit fly optimization algorithm (FOA) to improve the wind speed prediction accuracy. Next, the predicted values of the respective components are reconstructed to obtain the preliminary prediction values of wind speed. Finally, in consideration of the influence of atmospheric disturbance on wind speed, the Lorenz equation is introduced to correct the preliminary prediction values of wind speed to obtain the final wind speed prediction values. The high prediction accuracy and versatility of the prediction model of this paper are verified by experimenting with the wind speed data of two wind farms.