
Researchers at Radboud University have developed a new method to calculate the reliability of the power grid. This new method, based on Graph Neural Networks, is not only a thousand times faster but also more accurate than current methods. The results of the new method have been published in the journal Applied Energy.
The n-1 principle
Given the problems with grid capacity and contingency, the complexity of the power grid is increasing. Grid operators must ensure that the power grid remains reliable, even when a power cable fails. This is called the “n-1 principle”: In case of a failure, electricity must be able to be rerouted through alternative paths without causing problems.
During such rerouting, the load on alternative routes increases. Therefore, it is crucial to test whether these routes can handle the extra load. This involves checking not only the capacity of the cables but also whether the voltage and current and network stability remain within safe limits. Until now, for optimal results, the grid operators relied on mathematical calculations that checked all possible rerouting paths one by one—a process that could take hours.
The new approach
The new technology, developed by researcher Charlotte Cambier van Nooten and colleagues, uses machine learning. They have developed a “Graph Neural Network” (GNN) specifically adapted for power grids. This method views the entire network as a whole, rather than examining each route separately. Additionally, the method takes into account the properties of both the cables and the nodes in its calculations. The system learns to recognize patterns and works even for situations it has never encountered before.
Charlotte Cambier van Nooten states, “When there’s a failure, you want to quickly know the best method to solve it. Our new method can do this in seconds. Moreover, our method is on average 5% more accurate than traditional methods.”
The method has been tested on the medium-voltage grid, a complex cable network that delivers electricity between different substations. Grid operator Alliander has already begun implementing this new technology.
More information:
Charlotte Cambier van Nooten et al, Graph neural networks for assessing the reliability of the medium-voltage grid, Applied Energy (2025). DOI: 10.1016/j.apenergy.2025.125401
Citation:
New method enhances power grid reliability assessment (2025, February 24)
retrieved 24 February 2025
from
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
Leave a comment