Exploiting AI techniques to schedule maintenance for large-scale multi-component systems
Maintenance planning for complex systems has been considered challenging for a while now. In the paper ‘Artificial-intelligence-based maintenance scheduling for complex systems with multiple dependencies’ Van-Thai Nguyen, Phuc Do, and Alexandre Voisin from University de Lorraine elaborate on how AI techniques can be optimally used to schedule maintenance for large-scale multi-component systems, taking into account the impact of component dependencies.
To support this work, authors propose an AI-based framework to tackle maintenance decision-making in the case of unknown system cost models. The current paper focuses on modeling maintained systems with discrete state components, that suffer from stochastic and economic dependence as well as, on customizing MADRL algorithms to effectively optimize maintenance decisions of large-scale systems.
Last but not least, as AI-PROFICIENT partners mention, future work will focus on modeling methods that can integrate all three dependency types into maintenance models, and also improving the learning speed of MADRL algorithms.