New AI System Makes Machines More Efficient

Researchers at the University of Gävle have developed an AI system that can predict and plan the maintenance of industrial machinery. The system ensures that production does not need to stop unnecessarily, spare parts are replaced on time, and machines have a longer lifespa

One ongoing challenge for the industry is timing production with maintenance. It is crucial to schedule production stops during periods of lower demand or higher energy prices to maximize profitability. AI has become increasingly important, and today there are technical solutions that continuously monitor the status of machines to prevent unexpected breakdowns and stoppages.

A group of researchers at the University of Gävle has developed a new AI system that also allows machines to adjust their production speed and workload so that maintenance and spare part replacements coincide with production stops. With the help of this new technology, the stress on machines is reduced, and components do not need to be replaced unnecessarily.

"The technology enables the machines to indicate when they need maintenance and automatically adjust themselves to run at an optimal level, allowing maintenance to be predictable," says Niclas Björsell, scientific leader of the research area Intelligent Industry at the University of Gävle.

An example is a rolling mill with three rollers. Information from the AI system means that the entire production does not need to stop when one roller starts to wear out. Instead, the other two rollers can work harder for a period to compensate, allowing production management to plan a joint production stop and replace all the rollers simultaneously, thus avoiding three separate stops.

The AI system is primarily intended for use in the process industry, which is more sensitive to production stops than the manufacturing industry. The process industry includes companies that transform raw materials into products, such as the pulp and paper industry, steel industry, and mining.

"A stop in a machine has a greater impact in the process industry because the risk is higher that the entire production chain will break down during a stop. In the manufacturing industry, you can often have a buffer of products, whereas in the process industry, there is a continuous process where raw material is processed into the final product," says Niclas Björsell.

The next step is to test and implement the AI system in existing production.

"The new technology can be used in all automated manufacturing within the process industry, but also within the manufacturing industry. Above all, it benefits the utilization of facilities, as it is costly when machines stand still," says Niclas Björsell.

Facts

The research is led by Niclas Björsell, Professor of Electrical Engineering and scientific leader of the interdisciplinary research area Intelligent Industry.
The research has been conducted in collaboration with SSAB, Ovako, Alleima, and ABB.

The AI system was developed as part of a research project at the University of Gävle, which consists of four separate studies:

1. Björsell, N. and Dadash, A.H., 2021. Finite horizon degradation control of complex interconnected systems. IFAC-PapersOnLine, 54(1), pp.319-324. External link.
2. Hosseinzadeh Dadash, A. and Björsell, N., 2022, November. Adaptive Finite Horizon Degradation-Aware Regulator. In European Workshop on Advanced Control and Diagnosis (pp. 123-132). Springer Nature Switzerland. External link.
3. Dadash, A.H. and Björsell, N., 2023. Optimal Degradation-Aware Control Using Process-Controlled Sparse Bayesian Learning. Processes, 11(11), p.3229 External link..
4. Dadash, A.H. and Björsell, N., 2024. Infinite-Horizon Degradation Control Based on Optimization of Degradation-Aware Cost Function. Mathematics, 12(5), p.729. External link.