Solar Wind Speed Prediction With Two‐Dimensional Attention Mechanism

As more and high-technical systems are exposed to the space environment, extreme weather becomes a great threat human society. In solar system, is influenced by wind, such that reliable prediction of wind conditions in near-Earth environment effectively reduces impact on Solar speed improved making full use OMNI data measured at Lagrangian Point 1 (L1) National Aeronautics Space Administration (NASA) image observed Dynamics Observatory (SDO) satellite this work. Specifically, we propose model based “two-dimensional attention mechanism” (TDAM) predict speed. study, first analyze preprocess from 2011 2017. Second, considering characteristics time series data, adopt gated recurrent units (GRU) which can deal with long-term dependence as part our model. Third, design TDAM, enables network focus important parts. Three performance indices used: root-mean-square error (RMSE), mean absolute (MAE), correlation coefficient (CC). By comparing TDAM other models, find achieves best results, RMSE 62.8 km/s, MAE 47.8 CC 0.789 24 h advance. The experimental results show proposed improve accuracy