Home " Approach " Proof of concept chain project AI for Diagnostics and Maintenance

Proof of concept chain project AI for Diagnostics and Maintenance

Our society is increasingly dependent on modern complex software and hardware systems, ranging from microchip production lines and automated warehouses to hospital systems and our power grid. The cost of failure of such systems can be enormous, up to millions of dollars per hour for large process plants(link). A clear priority for the application of AI in the engineering industry is therefore intelligent diagnostics for improved reliability and for the maintenance of complex "cyber physical" systems. Since failing systems are expensive and risky, they must be made fully operational again as soon as possible, which requires specialized and well-trained personnel. The availability of such personnel is a challenge in itself(link). Rapid repair requires intelligent diagnostic and maintenance techniques, root cause analysis of failures, and guidance for service technicians.

Deploying AI for intelligent diagnostics

This project leverages the power of AI to increase system availability, reduce reliance on well-trained and experienced service technicians, while reducing cost-of-ownership and contributing to carbon reduction and sustainability. The project serves as a stepping stone to the next challenge of achieving "proactive/predictive maintenance. This project realizes two deliverables:

  1. A practical Intelligent Diagnostics methodology that supports and guides the service engineer
  2. Insight from actual business case(s) in which each partner in the value chain benefits.

Understanding maintenance prediction

Intelligent Diagnostics is an AI application that reuses and translates available system design information, such as system functionality, system decompositions and performance indicators, into diagnostic models. Combined with data-based algorithms, it determines the probability that a specific component failure has occurred. This approach uses probabilistic reasoning techniques to deal with probabilities and uncertainties.

In the chain project, use cases from multiple domains are deployed. The first use case is from ASML, VDL-ETG and Neways, in a second phase of the project this will be expanded with other use cases in various domains and company profiles (large companies and SMEs) all over the Netherlands. The implementation of this project will be done by the Embedded Systems Innovation group (TNO-ESI) as the lead agency in cooperation with the Technical Industry working group of Netherlands AI Coalition and all AI hubs. Translation of the results to the practice of the technical industry and for the creation of educational material will take place through a learning community to be developed.