sketch by Sven Winquist for the project

Doctoral project: Machine Learning in Manufacturing

Many manufacturing companies struggle with transitioning into the era of smart technologies, due to the fact that modern lab-grown methods and data intensive mega projects seldom fulfill expectations when applied in reality. This project aims to move research out of the laboratory and into reality, so that pragmatic methods can be developed.

Project information

Doctoral student
Felix Viberg
Welf Löwe
Assistant supervisor
Lars Håkansson
Participating organizations
Linnaeus University, SKF
SKF, The Knowledge Foundation
2021–1 Oct 2025
Computer and Information Science (Department of Computer Science and Media Technology, Faculty of Technology)

More about the project

Traditionally, manufacturing relied heavily on rugged machines and rugged people. This strategy proved successful in the era of materialistic abundance and a near infinite supply of working-class citizens.

But as time goes on – times change. Today, manufacturing relies on smartness and effectiveness, optimizing not only for a huge output to a minimal cost, but economic sustainability, delivery timing, eco-friendliness, workforce attraction and ergonomics. This change in mentality can be seen with the naked eye in most factories, due to that many of the old-era manufacturing lines are still being used.

Incorporating data intensive solutions, like machine learning and real-time big data analytics, into existing equipment requires bespoke implementations. However complicated, it is an exercise that requires research and cannot wait, as many companies struggle with this today.

With this project, SKF and Linnaeus University apply data intensive research at the lowest level of the organizational structure, to gain continuous real-world insight into the manufacturing operations. The project aims to adjust existing methods for machine inference so that these can be used directly, apply the new methods, and report the findings back to the scientific community.

It is expected that data intensive solutions trade off better results to higher computational cost, a relationship that can only be fully grasped after finalizing real implementations. The project also aims to develop new ways of visualizing the result of analysis as new data emerge.

The project is part of the research in the Data Intensive Software Technologies and Applications (DISTA) and Linnaeus University Centre for Data Intensive Sciences and Applications (DISA) research groups.