IMPROVING SOLAR ENERGETIC PARTICLE PREDICTION WITH MACHINE LEARNING

Abstract

This study advances Solar Energetic Particle (SEP) prediction through advanced machine learning techniques, mitigating the risks SEPs pose to space missions, satellites, and terrestrial systems. Using historical and real-time data from NASA and ESA, models based on Long Short-Term Memory (LSTM) networks, Convolutional Neural Networks (CNNs), and Random Forest algorithms were developed. Careful dataset preparation, hyperparameter tuning, and cross-validation ensured robust model performance. Among these, CNNs demonstrated superior accuracy and precision, making them a valuable tool for SEP forecasting. Overall, this work enhances machine learning capabilities for space weather prediction, contributing to safer and more reliable space operations