An optimized asset management petri net model for railway sections
Ali Saleh · Materials research proceedings · 2023
Abstract. Railways are important ways of transportation that are used massively. This makes it important to create an optimized asset management model that helps in reducing its Operation and Maintenance (O&M) costs while maintaining the quality of service and safety. Reinforcement learning (RL) is an adequate model for optimizing decisions based on unrelated factors as it connects the decision to a final goal without understanding the problem details. Also, it allows for automatic policy updates without any user intervention. On the other hand, the Petri net (PN) model, which is a bipartite graph of transitions and places, are adequate to be combined with Reinforcement learning since RL actions can be directly described by the PN transitions. In addition, PNs are suitable for maintenance modeling since it can model heterogeneous information, parallel operations, and synchronization, and provide a graphical interpretation. In this study, the Petri net method is used with Reinforcement Learning to create a tool for modeling and optimizing decisions within the maintenance of railway sections while taking into account several factors.