Leonie Meldt from the Institute for Production Management, Technology and Machine Tools (PTW) explores this question at the IEEE International Conference on Recent Advances in Systems Science and Engineering 2025 in Singapore, where the paper was honored with the Best Paper Award.

Co-authored by Anh Minh Augustino Doan, Matthias Weigold and Joachim Metternich, the paper addresses a key challenge in industrial scheduling: while RL enables reactive rescheduling, many approaches still lack mechanisms to proactively prepare for disruptions.

Key Takeaways and Benefits:

• Proactive resilience through controlled disruption training
  Chaos Engineering enables synthetic disruptions during RL training, so the agent learns how to handle failures and turbulence before they occur in real operations.
• A modular microservice architecture for resilient APS systems
  The proposed architecture defines dedicated services for optimization (RL scheduling), disruption management (Chaos Manager), consistency checking, and user interaction, enabling scalable and maintainable implementation

Read the full paper here: IEEE Xplore (wird in neuem Tab geöffnet)