The challenge

The more electrified society becomes, the more dependent we become on the power system working properly. And the more complicated the power system becomes, the more vulnerable it is. Small faults that turn into major faults can have serious consequences for society, both practical and financial.

The goal

In EarlyWarn, the goal is to develop a monitoring system, driven by big data and domain knowledge, that detects and identifies errors, preferably several minutes before they occur.

The project

The project uses large amounts of measurement data, which is analysed using machine learning models that in turn develop a predictive algorithm. The theory is that small variations (fault signatures) in the voltage in the grid can predict breakdowns in components and other events that could lead to disruptions and interruptions in the power system. Preliminary trials show good results, although better for some fault categories (total operational interruptions) than for others (earth faults and voltage dips).

Project team members

  • SINTEF Energy

  • SINTEF ICT

  • The Norwegian University of Science and Technology (NTNU)

  • University of Strathclyde

  • Statnett

  • Nettalliansen

  • NTE Nett

  • Lyse Elnett

Funding

  • The Research Council of Norway (RCN)

  • Statnett

  • Nettalliansen

  • NTE Nett

  • Lyse Elnett

You can also read about the project at:

https://www.sintef.no/en/projects/earlywarn/

https://blog.sintef.com/sintefenergy/machine-learning-can-predict-faults-and-disturbances-in-the-power-system/