Industrial doctoral student's AI research improves industrial maintenance work
After five years as an industrial doctoral student at Mid Sweden University and SCA, Adam Lycksam is ready to present his thesis. With a focus on machine learning and predictive maintenance, he has developed methods that will predict faults in machines – and thus prevent costly downtime in industry.
Unplanned downtime is an expensive challenge for industry. Therefore, during his time as an industrial doctoral student at SCA, Adam Lycksam has focused on developing new AI-based methods to predict and prevent failures in technical equipment.
"The goal has been to create methods that really work in the complex industrial environment. It has required a combination of advanced machine learning and an understanding of processes and machines in the factory," Adam describes.
His research is based on three central parts: identifying faults, predicting lifespan – and building an AI-based framework for predictive maintenance that can address the challenges that exist in the industry.
One of the contributions in the thesis is how prior knowledge can be integrated into machine learning systems to improve both precision and reliability.
"A great insight is the value of using the knowledge we already have about machines and faults when we train our models. It is often not optimal to simply enter data into an AI system. If we combine deep learning with domain knowledge, we will get solutions that are more robust. Here, research and industry could collaborate more," says Adam.
He also emphasizes that it is important to use several types of data sources and hybrid models – where different methods interact and thus create more reliable results.
The world of research meets the reality of industry
Being an industrial doctoral student has given Adam a unique position where he has been able to move between the world of research and the reality of industry. This has made it possible to test methods in real environments – something that he emphasizes is extremely important for research to contribute to methods and tools that are relevant and deliver what the industry needs.
"It's only when you test in real life that you see which solutions actually work. The collaboration with colleagues at SCA and supervisors from both the university and the company has been crucial to being able to get to where we are today," he says.
Adam's supervisor at Mid Sweden University has been Professor Mattias O'Nils, and from the industry side, Jenny Klöver, Group Manager for IT and Operational Technology at SCA, has had a key role. She sees great benefit in having an industrial doctoral student attached to the company.
"It has been extremely valuable to have an industrial doctoral student! In addition to the proximity to the activities, and the opportunity to understand them in the best possible way, they have academia with the network of both researchers and other doctoral students to help them. We have not had the resources that have been able to work on these issues in such a focused way as Adam has. The results we have received will be implemented at several of our facilities and also developed further. This has created added value for SCA as a whole," she says.
Research for several contexts
Adam is now looking forward to continuing to develop the methods in his new role at the Central Digitalization Support unit within the SCA Group.
"My starting point has always been that there should be an opportunity to use my research and that it should be able to be used in several contexts. It is important to build up knowledge and share it. In my new role at SCA, I will now continue to develop the methods together with my colleagues," he says.
Adam is also part of the industrial doctoral school Smart Industry Sweden, where Mid Sweden University, Halmstad University, the University of Skövde, University West and Jönköping University collaborate with various companies by employing industrial doctoral students. The initiative is financed with the support of the Knowledge Foundation.
Join us for the public defense of the doctoral thesis!
On 16 June, Adam Lycksam will present his thesis.
Time: 13.00.
Location: Campus Sundsvall, room C312.
The presentation can also be followed via Zoom and Youtube.