AI algorithms to predict problems before they occur

Vast numbers of sensors monitor industrial production systems and raise an alarm when they detect faults or deviations. A Vinnova-financed project is to explore how artificial intelligence can be used to detect patterns in the deviations and prevent problems before they arise. 

Sensors monitor industrial production systems AI algorithms can be used to detect patterns in deviations and faults.  sergeyryzhov

“There is a huge potential in combining the knowledge we have within artificial intelligence, AI, and life-cycle engineering to improve not only the environment but also the economic performance of companies”, says Tomohiko Sakao, associate professor in the Division of Environmental Technology and Management.

Tomohiko Sakao is coordinator for the three-year project Adapt 2030, financed by Vinnova within the strategic innovation programme “Produktion2030”. Vinnova is contributing more than SEK 9 million. The overall aim of the project is that research should contribute to making Swedish manufacturing industry more competitive by 2030, and here AI is an important part.

Detecting patterns

Large amounts of data have been collected about faults and deviations in industrial production systems, including deviations during manufacture and assembly. The data also contain information about faults that arise when products are used and maintained.

Tomohiko Sakao Photo credit Monica Westman“Several digital technologies are available that make it possible to provide feedback – about faults and deviations during manufacture, assembly, use and maintenance – during the complete lifetime of a product to those who carry out design and construction”, says Tomohiko Sakao.

The idea now is to use AI algorithms to detect patterns in the faults and deviations such that it is possible to define the factors that cause faults, and in this way prevent them arising or react to them faster. It may be possible to quickly change a critical detail in the design or construction. A product can also behave in different ways depending on where it is used: in damp, sandy or windy conditions, for example, and indoors or outdoors. The researchers hope that it will be possible for the design and manufacture to be adapted to how and where the product is to be used.

“A few articles from research in North America have recently reported positive results from this type of work, and we plan to produce a proof-of-concept during the next three years, which we can use to confirm that the principle works, not just in theory but also in practice. This is both a step towards a more circular economy and an opportunity for AI to show its worth for companies in practice”, says Tomohiko Sakao.


The full name of Adapt 2030 is “Adaptive lifecycle design by applying digitalization and AI techniques to production”. In addition to researchers at LiU, researchers from Mälardalen University are participating, and others from large companies such as Volvo Construction Equipment and Siemens Industrial Turbomachinery, AI companies such as Addiva and Imagimob, the consultancy Semcon, and the National Road and Transport Research Institute VTI.
The project is scheduled to end in February 2023.

Translated by George Farrants

Latest news from LiUShow/Hide content