 |
Prophecy Prognostic Model Selection for Predictive Maintenance and an Integrated Reinforcement Learning-based Production Scheduling in Dynamic Manufacturing Systems This research project is a collaboration between the Bremer Institut für Produktion und Logistik (BIBA) at the University of Bremen, Germany, the Federal University of Rio Grande do Sul (UFRGS), Brazil, the Federal University of Santa Catarina (UFSC), Brazil, and the Federal University of Amazonas (UFAM), Brazil. The goal of this project is to develop a self-adaptive model selection method for predictive maintenance that is fully integrated into production and maintenance planning. To achieve this, a machine learning-based approach will be developed, enabling the automated selection of suitable prognostic models for different system configurations and conditions. A key aspect is the incorporation of reinforcement learning for the dynamic optimization of machine availability and utilization in real time. This is based on a digital representation of the production system, which allows for the evaluation of decision impacts using production logistics KPIs. This performance assessment enables targeted feedback between meta-learning and reinforcement learning, contributing to the continuous improvement of the system. A key aspect of the project is the integration of reinforcement learning to dynamically optimize machine availability and utilization in real time. This is based on a digital representation of the production system, allowing the assessment of decision impacts using production logistics KPIs. The continuous feedback loop between meta-learning and reinforcement learning facilitates the ongoing improvement of the system. To validate the developed methods, a simulation-based environment will be created, which replicates the relevant production and maintenance processes with the required level of abstraction. Finally, the developed system will be tested in two industrial use cases in Germany and Brazil. Contact persons: H. Engbers   (Project manager)R. Caballero Gonzalez Funded by: DFG Duration: 01.02.2025 - 31.01.2027 See project's publications List all projects |
