ABSTRACT

COLLABORATION BOOSTS DIGITALIZATION - HOW EUROPEAN ELECTRICITY GRID OPERATORS SHARE DATA TO TRAIN BETTER AI-MODELS.

At EuroMaintenance 2024, we would like to present an innovative project jointly undertaken by six leading European grid operators: Elia Group/50 Hertz, E.ON/E.DIS, Swiss Grid, SBB, Amprion and APG. Launched just a year ago, this project uses artificial intelligence (AI) to improve the maintenance of electrical assets, with a particular focus on optimizing the eSiciency of pole and line maintenance. This project highlights the significant potential of collaborative innovation as it addresses a common challenge faced by individual grid operators - the inability to singe handedly train an AI model due to the low number of images of damaged assets resulting from their zero-fault strategies.

The collaborative was born from a common motivation: across Europe, there's a widespread recognition that the energy transition requires greater eSiciency in the maintenance of electricity networks, driven by an increasing demand for electricity and the resulting need for network expansion. Traditional manual inspection processes, in which human workers review individual images captured by drones, are not only time-consuming, but also impractical given the scale of operations. Therefore, the central question guiding our project was: how can we make current transmission line maintenance and repair processes more efficient?

Our answer to that question consists of a comprehensive shared database of images of damaged and undamaged assets taken by helicopter or drone. This image database was used to train AI models for asset and damage detection. The final AI engine will be able to automatically identify individual transmission line components and potential damage, as well as the type of damage.

The project was carried out under the umbrella of our European Ecosystem uniting seven grid operators, with Infront as the intermediary. The goal is to join forces to deliver faster, better and more cost-eSective innovations. The seventh ecosystem member which was not an active member of the shared database project, Red Elèctrica, contributed their already existing AI model that was further trained by a software developer with the partners' images, saving 2-3 years of development time.
By pooling resources, the project reduced developmental costs by 75% and achieved significant milestones in just three months, including the development of 17 highly accurate AI models, all of which were combined into one eSective AI engine with precision/recall rates ranging from 0.85 to 0.95, implying very accurate damage detection. Experiments showed that a collaboratively trained AI engine significantly outperformed an AI engine trained with only one partner's image database by more than 25%. Additionally, we successfully build a shared image database of over 6000 images with over 1000 labeled images.

For one partner alone, the AI solution could save up to 2,500 working-days of manual inspection, dramatically reducing costs and helping to address a labor shortage that is expected to grow in the future. We are more than happy that the success of the project was recently recognized by the Maintainer Award for innovation.

Our project demonstrates the potential that lies in the collaborative digitization of maintenance. It highlights not only the need to share data, but also the direct positive impact on the speed, eSiciency, safety and cost-eSectiveness of maintenance applications.

BIO

Thomas Sindemann is co-founder and partner of Infront Consulting and has also been Vice President of KPS AG since 2018.
"Transformation succeeds when the right strategy meets excellent implementation". As a pragmatic strategist, Thomas Sindemann develops and implements diverse, mostly digital transformation and innovation programmes. His focus is on identifying cross-industry potential. By intelligently merging companies with similar problems, he helps them to explore technologies faster through real synergies, which significantly strengthens their growth power in the market. Effective change in a short time, together with the customer and for noticeably more success - that is what drives him.