Will we be able to chat with our own data?

New technical solutions can become obsolete in a few months. A challenge when an entire factory is digitizing its production and looking for sustainable digital solutions that will last a long time.

This week, our Digital Factory Leader, Perparim Sylejmani, held a presentation on Nexans' Digital transformation at the Maintenance Fair in Gothenburg, where he and our partner Wilhelm Söderkvist Vermelin from RISE described how Nexans has worked within the Dfusion project since 2021.

From the beginning, Dfusion was a project that, by measuring and analyzing machine data from many different sources, wanted to be able to predict the conditions under which possible breakdowns could occur – Predictive Maintenance. At Nexans they have succeeded – but over the past three years or so, the original objective has partly changed.

– What we see is this: We need Conditional Based Maintenance. That is, we don't just want to be able to predict the future; we need to be able to act in real time. The data should be able to give us a red or green light immediately, says Perparim Sylejmani.

Wilhelm Söderkvist Vermelin at RISE has helped Nexans to analyze the data. The challenge lies both in finding technical solutions that make it possible to extract the data, but also in knowing what to look for in the gigantic amount of information.

To collect and own its data, Nexans has installed a server that can store both production data and all other data. A server where everything is gathered in a system. Again, the project has seen that the premises and goals that applied when it started in 2021 are no longer sufficient.

– When we started, Chat GPT did not exist. But what we see now is that maintenance technicians or operators need an easy way to ask for help with the analysis. They don't have time to analyze the data themselves – it's too much data – and too complicated. We need a communication tool – someone to ask, says Wilhelm.

– Initially, we thought that the analysis would result in a report to management, but Nexans is now building its own Chat tool (Large Language Model) with our data. Ultimately, we hope that the operator will only be able to ask: Can you see historically that a similar deviation occurs? On which machine? What did we do then? What do you recommend we do, says Perparim.

Tobias Karlsson works in the Maintenance Department as a maintenance technician at Nexans, where the entire department for preventive maintenance has had to change working methods and learn new things. He also agrees that the new way of working has challenges.

– It is so huge. For me, the most difficult thing is to see the logic sometimes and comprehend how it works, he says.

As always – despite AI, fantastic technical solutions and high tech – employees are central to this transition. A major task in digitization has been to map the skills that Nexans employees have today, the skills that will be needed in the future and how that gap can be solved. Here too, the rapid development makes it difficult to know whether the investments we make are the right ones.

– I foresee that the balance will be tough. Not least because we are a Group. I think the challenge going forward will be to know when to invest and when to slow down. To rethink quickly because it is required – and not to think that the original plan will work 100 percent, and to make everyone understand that, says Perparim Sylejmani.

About Dfusion
Dfusion – Data fusion of production disturbance data is a research project funded by Vinnova. Almost 50 percent of the Swedish production capacity is unutilized due to disturbances caused by, among other things, process deviations. The purpose of the project is to support companies in disruption management and create more efficient and sustainable production systems.