Microsoft Research develops machine learning system to forecast space-weather risks for U.S. power grids

Microsoft Research describes an end-to-end machine learning pipeline that forecasts space-weather risks for 66,935 substations across the continental United States. The system uses real-time solar-wind measurements from the L1 Lagrange point to generate forecasts of Auroral Electrojet (AE) and Disturbance Storm Time (Dst) indices, then combines these with local factors such as latitude, geology, and ground conductivity to produce location-specific risk estimates.

During the 2020–2026 evaluation period, the AE predictor captured nearly the full observed range of AE activity and outperformed several empirical solar-wind-based approaches. The Dst predictor outperformed the Burton equation on 62.2% of individual hours during the most geomagnetically active periods and provided a wider prediction range than Burton-style baselines.

In the final risk stage, the system estimates dB/dt—the rate of magnetic-field change associated with geomagnetically induced currents—and translates these predictions into location-specific risk estimates. It achieved detection rates of 76.5% for major events (≥10 nT/min), 81.2% for severe events (≥20 nT/min), and 64.1% for extreme events (≥50 nT/min), with higher detection rates at northern stations where geomagnetic activity is strongest.

The pipeline was developed using only public data sources, including NASA OMNI, Kyoto World Data Center, INTERMAGNET, U.S. Geological Survey magnetometer observations, and GridSFM-derived grid data. A system of 50 AI agents assisted in exploring features, validation strategies, and model configurations across the pipeline.

The work was conducted during a summer internship and published by Microsoft Research on September 30, 2026. The post highlights the challenge of forecasting space-weather impacts on critical infrastructure and notes that the system aims to provide grid operators with enough warning—30 to 60 minutes—to take protective actions before a storm arrives.