• 3 min read
GridFusionX cuts power reserve costs by up to 66%
GridFusionX improved power forecasts by up to 56% and cut reserve costs by as much as 66% across 10 European regions.
Image: TechXplore
The simulation lab at the Center for Advanced Power Systems. Credit: Scott Holstein/FAMU-FSU College of Engineering
A forecasting system developed by researchers at the FAMU-FSU College of Engineering and Florida State University’s Center for Advanced Power Systems improved electricity predictions by up to 56% and reduced reserve costs by as much as 66% in tests across 10 European regions.
Called GridFusionX, the system is designed to help grid operators balance electricity supply and demand as renewable generation adds uncertainty to power planning. The research was published in IEEE Transactions on Network Science and Engineering in 2026.
How GridFusionX forecasts power demand
Grid operators must decide how much reserve electricity to keep available. Holding too much reserve increases operating costs, while holding too little can leave the system exposed to demand spikes or sudden drops in renewable generation. GridFusionX aims to narrow that uncertainty by generating both forecasts and confidence intervals.
Recommended reading
Gemini Spark can now book through Chrome
The model combines multiple data sources, including:
- Historical electricity demand
- Renewable energy generation
- Energy market prices
- Other regional power-system data
Its central technique is a graph neural network, which represents connected entities as nodes and models the relationships between them. In GridFusionX, the network captures how conditions in one area can affect neighboring regions through shared transmission lines, weather patterns and energy markets.
That network-aware design distinguishes the system from forecasting approaches that treat regions more independently. The model produces spatially connected predictions while also indicating how uncertain those predictions are, giving operators information they can use when setting reserve levels.
“Our research looks at smart systems as a dynamic puzzle: every factor, from different energy sources to shifting demands in cities and changing prices, and how they fit together. What happens in one area can affect neighboring areas because they share power lines, weather patterns and energy markets. By continuously analyzing these pieces, we can predict energy needs more accurately and adapt to ensure both reliability and cost savings.”
Accuracy and reserve-cost results
In the reported tests, GridFusionX improved forecasting accuracy by up to 56% and lowered reserve costs by as much as 66%, while maintaining reliable service. The system is intended to help utilities prepare more efficiently for changes such as sudden demand increases or reduced renewable output.
The researchers describe the approach as multimodal because it combines several types of information rather than relying on a single historical series. Associate Professor Olugbenga Moses Anubi, a study co-author, said reducing uncertainty can support more adaptive decisions in areas ranging from electricity use to traffic and weather prediction.
“Right now, utility companies estimate your usage to make sure you never underpay, which almost always means you overpay. With our approach, predictions become more precise, so your bill matches what you truly use.”
The source does not provide a commercial release date, utility deployment schedule or pricing for GridFusionX. It also reports the performance results from the researchers' tests without identifying a separate third-party validation of the benchmark figures.
Research and student training
The project was led by doctoral student researcher Quoc Bao Phan and involved four faculty members. FAMU-FSU faculty are incorporating related concepts into courses covering cybersecurity systems for electric grids and artificial intelligence for power systems.
Associate Professor Ravikumar Gelli described the project’s goal as “engineering intelligence” to help utilities balance generation and load while reducing costs for consumers. Anubi said the collaborative structure also gives students experience working with several faculty members instead of a single adviser.
The full study is titled GridFusionX: Network-Aware Probabilistic Forecasting for Multi-Regional Power Systems. Its DOI is 10.1109/tnse.2026.3678493.
Gaby Clark — MA in English and copy editor since 2021, with experience in higher education and health content.
Robert Egan — bachelor’s degree in mathematical biology and master’s degree in creative writing.
AI Editor
Ava covers the rapidly evolving world of artificial intelligence, from foundational models and research labs to the real-world economics of intelligence. With a background in computational linguistics, she cuts through the hype to find out what actually works. She firmly believes that benchmarks are just marketing until reproduced in the wild.
via TechXplore
