Predictive maintenance

How a neuronal algorithm protects windmills

Can computers and algorithms be used to help the maintenance of windmills? A team of experts from HanseWerk AG (one of the four German distribution system operators by E.ON) and data scientists of the E.ON DataLab developed a self-learning algorithm that predicts when medium voltage cables in windmills need to be replaced. The algorithm is self-learning and the computing systems are inspired by the same biological neural networks that constitute animal brains. This is called a neuronal algorithm. As far as we know, we are the first energy company that uses "artificial intelligence" (AI) that prioritizes replacement measures. This project sets the example for grid operators worldwide. Our self-learning system uses all sorts of data in order to identify behavior patterns of the electricity production and spots inconsistencies. Next to our existing data (for example geo information data, asset data, grid operations), the computing system also considers external data (for example weather, lightning and salt content data) to identify behaviour patterns. We can use our solution with a minimum required database, and this allows our system to automatically learn on a continuous basis. This innovative system proves to be succesful. Research has shown that the intelligent approach is good for an improvement of 30% compared to the conventional approach. An additional advantage: procedures for replacement measures have become more comprehensible and transparent. On top of that, we're able to distribute maintenance-budgets in a far more efficient way.

Further developments

We continue to improve our neuronal network. First of all, we've just started to implement our product on a bigger scale, going from an experimental to an industrialized phase. Second of all we've finalized the analytical model for medium voltage cables and substations with real-time data from network control centres. This way we can improve the precision of our predictions even further. Finally, we'll extend our analytical model to also cover other assets, such as low voltage cables.

Of course, we also want to share these results. That's why we put a lot of time in informing control centres and project and municipal managers of E.ON. Our innovative and user-friendly visualization tool helps them benefit from an innovative comparative advantage when they talk with our partners.

The algorithm for a better future

An innovative project like this is never finished. That's why we continue to irmprove our neuronal algorithm. We continue to build on a significant and sustainable "step change" for replacing cables in our grid. To promote E.ON as an innovative grid operator, we adopt and industrialize the developed model to the distribution system operators in our company.

Thanks to this project, we position ourselves as pioneers in artificial intelligence and as an innovative grid operator. The results: better security of supply and better network quality for our network customers and partners. We believe that innovative technology and applications are the future and that they will give us a big advantage over our competition.