Research on Prediction of Geomagnetic Kp Index Based on ConvGRU Hybrid Model

Magnetic storms are intense global geomagnetic disturbances that can have significant impacts on human society. The Kp index shows how strong magnetic storms are. Predicting it is really important for studying geomagnetic storms. With the rapid development of deep learning, neural networks have shown great potential in predicting geomagnetic storms. This study builds a ConvGRU hybrid neural network model. It takes solar wind parameters as input to model and predict the Kp index. When we look at the CNN and GRU models, the ConvGRU hybrid model has an RMSE of 0 for its forecast error.498583, and the Pearson correlation coefficient is 0.904571. Compared with the single CNN and GRU models, the forecast effect of the ConvGRU hybrid model is better.