Comparative Analysis of Deep Learning Models for Solar Flare Prediction in Space Weather Forecasting

This research looks at the efficacy of a hybrid LSTM and GRU model that was trained on data from 2002 to 2016 in order to estimate the categories of solar flares. The model's performance is evaluated using metrics sxuch as accuracy, precision, recall, and F1 score on both test and new datasets. Additionally, the impact of training parameters like as epochs, batch size, and recurrent dropout is examined. The results illustrate the prediction accuracy of the model for solar flare categories and point to the technique's potential use in solar activity forecasting.