Preventing malfunctions with artificial intelligence
NWO has rewarded the ZORRO project: Engineering for Zero Downtime in Cyber-Physical Systems via Intelligent Diagnostics. This grant comes from the call Next Generation High-tech-Equipment: cyber-physical systems (KIC).
Malfunctions. We all suffer from them: payment apps that don't work, websites that are down, or software that doesn't cooperate. Often just irritating, but for businesses, downtime is a huge expense whose bill can run into millions. Some malfunctions can spread like an oil slick: we saw this, for example, in the power malfunction in Flevoland where the failure of a fuse led to a short circuit. Then a fire started, trains did not run, and locks were unable to function.
"We want to achieve breakthroughs in complexity with ZORRO"
To prevent such problems, the University of Twente, TNO-ESI, Saxion University of Applied Sciences, and Vrije Universiteit Amsterdam, are working on using artificial intelligence to better predict and prevent malfunctions. They have received a three million euro research grant for the project ZORRO (Engineering for Zero Downtime in Cyber-Physical Systems via Intelligent Diagnostic).
The project is headed by Twente University Professor Marielle Stoelinga, and Co-Project Leader Carmen Bratosin of TNO-ESI. Industry participants are ASML, Canon Production Printing, ITEC, Philips and Thermo Fisher Scientific. The team is receiving a financial boost from NWO, from the so-called KIC program.
The ZORRO project is working on diagnostic methods for high-tech systems, such as MRI scanners and printers. By continuously monitoring their behaviour with appropriate sensors, algorithms from AI can detect anomalous patterns in the sensor signals, and relate these to their root causes. Appropriate measures, such as replacement or repair, can then be taken in time to prevent malfunctions.
"We want to achieve breakthroughs in complexity with ZORRO," explains Mariëlle Stoelinga, professor of Risk Management for High-tech Systems at the University of Twente. "Then it's not about diagnostics for simple components, but for whole systems; efficient monitoring systems through smart combinations of sensors; automatic diagnostics by capturing domain knowledge in diagnostic models and integrating it in the development process for high-tech systems."
Co-project leader Carmen Bratosin of TNO-ESI is very pleased with this success.
This multi-company programme gives us confidence in broad application. It allows us to accelerate our efforts to realise ESI's diagnostics roadmap for the high-tech industry
The researchers will receive a contribution from the call Next Generation Hightech Equipment: cyber-physical systems, which is part of the Knowledge and Innovation Communities (KIC) research programme that is intended for groundbreaking innovative solutions that have social and economic impact. It involves companies, knowledge institutions and public authorities co-investing in the business application of knowledge to tackle major societal challenges with smart technologies.
NWO has awarded two projects that aim to improve the performance of high-tech systems, in particular cyber-physical systems. Within the projects, knowledge institutions are working together with industrial partners on reliable high-tech motion systems for lithography machines, for example. PPP research is also starting to establish diagnostic methods for complete high-tech systems, such as MRI scanners.
The grants come from the call Next Generation High-tech-Equipment: cyber-physical systems (KIC). The call aims to work on multidisciplinary integration of cyber-physical systems. The physical and digital components of a cyber-physical system are inseparable. By improving the interaction between the two components, progress is possible in areas such as efficiency, affordability and predictability. The Netherlands is an international leader in ultra-precise high-tech equipment. As such, the Dutch high-tech industry makes an important contribution to the prosperity and earning power of the Netherlands.