Symh index prediction with Neural Basis Expansion Analysis for Time Series (N-BEATS)

The Geomagnetic SYMH index is commonly used to measure disturbances in geomagnetic activity, such as the impact on ground-based technological systems resulting from Sun-Earth interactions. This measure can help mitigate potential damage and disruptions caused by space weather events. Recently, artificial intelligence (AI) has garnered increasing attention for its capabilities in predicting tasks, particularly due to its advantages in analyzing large datasets. Significant advancements in various model architectures for predicting the SYMH index have emerged, including empirical methods, machine learning, and deep learning techniques. However, challenges persist in this research area, as accurately predicting the SYM-H index remains difficult due to the dynamic nature of geomagnetic data. In this work, a new deep learning model of Neural Basis Expansion Analysis for Time Series (N-BEATS), which utilizes high temporal resolution data of one-minute SYMH index readings from the peak of most recent solar cycles (specifically, solar cycle 25). Our findings indicate that this new model has significant potential in capturing the temporal patterns of the SYMH index, achieving prediction accuracy of approximately 99%.

Text 121876.pdf - Published Version Restricted to Repository staff only Download (733kB) | Request a copy

Actions (login required)

View Item