Multi-Step Wind Power Prediction Method Based on Bi-GRU and Spatial Attention

Aiming at the prevailing problems in the wind power industry, such as the volatility of wind power and the stability of the power grid system. This paper effectively combines the BI-GRU (Bidirectional Gated Recurrent Unit) algorithm with spatial attention and proposes a multi-step wind power prediction network structure independent of meteorological data. Through the analysis of a large number of experimental data of this model, it is proved that the accuracy of wind power prediction is greatly improved and the response speed is faster, which provides a strong support for the reliability analysis of wind power prediction in actual scenarios.