EMD-GRU Based Wind Power Prediction Model

Abstract

A proposed wind power prediction method that relies on the empirical mode decomposition algorithm and gated recurrent unit neural network aims to achieve increased safety and stability of wind power grid connection while alleviating the pressure of peak and frequency regulation on the power system. Using the historical wind power as input, the data is first decomposed using the empirical mode decomposition algorithm, and then the gated recurrent unit neural network is used to explore the connections between the data to form the prediction feature vectors, which are then fed into the neural network for training, and the results of each component of the wind power prediction are output, and finally summed to obtain the final wind power prediction results. The results of predicting the magnitude of output power of wind turbines in this way on a European coast show that the proposed prediction method has significant advantages in terms of prediction accuracy compared with common gated recurrent unit neural networks, back propagation neural networks, and artificial neural networks.