Ultra-Short-Term Wind Power Forecasting Based on VMD-GRU-Transformer

Abstract

Accurate wind power prediction is essential for the stable operation of power systems. Aiming at the problem of insufficient accuracy of ultra-short-term wind power prediction, a combined prediction model based on VMD-GRU-Transformer is proposed. Variational Mode Decomposition (VMD) is used to split the wind power data into different intrinsic mode functions (IMFs) to weaken the non-stationarity of the original series. The combined GRU-Transformer network structure is designed to utilize gated recurrent unit (GRU) instead of the original word embedding and positional coding links, and feature fusion is performed on the input data to fill in the gaps in Transformer where the relevant information is not fully considered. Relying on the self-attention mechanism in Transformer to capture the time dependence of sequence data for prediction. Finally, a case analysis is performed with a public dataset, and the results show that the proposed model has higher prediction accuracy compared to other existing models.