Prediction of Wind power rolling interval based on VMD-PSR-LSTM-QR

Abstract

The integration of wind power into the power grid brings great challenges to the safe operation of the power system. Predicting the wind power interval can avoid effectively the randomness and volatility of the integration, and protect the security and stability of the power grid. Considering the non-stationary characteristics of wind power, this paper proposes a new wind power rolling interval prediction method based on VMD-PSR-LSTM-QR. First, wind power is decomposed into several subsequences by variational mode decomposition (VMD). Second, phase space reconstruction (PSR) maps the subsequences to a high-dimensional space and constructs them. Then, long short-term memory (LSTM) networks are used for training different subsequences, and quantile regression (QR) parameters are determined. Finally, the predicted results of each subsequence are reconstructed. This paper uses actual wind power data in China as examples to verify the proposed method, and it is found that the proposed method achieves significantly higher prediction accuracy than traditional methods.