Advancing Space Weather Forecasting: A Comparative Analysis of AI Techniques for Predicting Geomagnetic Storms

Abstract

Forecasting geomagnetic storms is crucial for mitigating their potential impacts on technology and infrastructure. This research explores the use of artificial intelligence (AI) techniques, particularly linear regression, and Long Short-Term Memory (LSTM) networks, for predicting geomagnetic storms using the OMNI dataset. The dataset, comprising various solar and interplanetary parameters, was preprocessed by scaling features and removing null values. A linear regression model achieved a Root Mean Squared Error (RMSE) of 5.95 and an R² score of 0.77. In contrast, the LSTM model, designed to capture temporal dependencies, significantly outperformed linear regression with an RMSE of 1.46 and an R² score of 0.99. These results demonstrate the potential of LSTM networks in accurately forecasting geomagnetic activity, thus providing a valuable tool for space weather prediction and the protection of critical technological systems.