Fujitsu and Tokai University System advance AI-driven space weather prediction

Fujitsu and the Tokai National Higher Education and Research System (THERS) have jointly developed a new AI-based space weather prediction technology that improves the forecasting of solar radiation events and their potential impacts. The innovation combines probability estimation with the identification of past, comparable events using Wide Learning, Fujitsu’s explainable AI (XAI) technology within the Fujitsu Kozuchi portfolio.

Moving beyond flare magnitude alone
Solar radiation events have traditionally been difficult to predict because their severity does not always correlate directly with solar flare magnitude. Even relatively small flares can deliver dangerous radiation doses in space. By applying AI, the new approach uncovers complex causal relationships behind these events and presents them in an explainable format, allowing operators to understand why a specific risk level is predicted.

Learning from past events
A key feature of the technology is its ability to automatically identify historical solar radiation events that closely resemble current predictions. By presenting these analogous cases—along with their documented impacts and countermeasures—the system provides practical, evidence-based insight to support operational decisions.

This capability is particularly valuable for:

Toward resilient space and terrestrial infrastructure
Looking ahead, Fujitsu and THERS plan to further enhance the technology to help protect power grids, satellite communications, GPS systems, and polar aviation routes from space weather disruptions. The collaboration demonstrates how explainable AI can serve as a foundational tool for building a more sustainable and resilient society, extending benefits from space exploration to everyday critical infrastructure on Earth.

www.fujitsu.com