Wind power prediction based on SSA-VMD-GRU hybrid model

Accurate wind prediction is essential to enhance wind energy and operational efficiency of wind farms. Implementation of combined prediction models improve prediction accuracy. This research introduces an improved model for the combined prediction of wind power, SSA_VMD_GRU, which integrates the ability of Singular Spectrum Analysis (SSA) in principal component extraction and the advantage that Variational Modal Decomposition (VMD) can effectively extract components of different frequencies in the signal to do data preprocessing. Ultimately, Gated Recurrent Unit (GRU) is combined to form a comprehensive prediction model. Comparison of the multi-class combined prediction models and practical examples show that, by comparison, the combined model developed in this research can deeply explore the potential law of wind power change and provide more accurate prediction capability, which can provide a reference for scheduling management and operation optimization of wind farms.