Maintenance planning for bridges using hierarchical reinforcement learning

This paper presents a hierarchical deep RL framework for maintenance planning on bridges. The proposed HRL framework provides advantages in scalability, and interpretability by allowing to visualize the decision boundaries of policies. The RL environment in this study is based on state-space models (SSM), which enables including the deterioration speed alongside the condition in the decision-making analyses. The performance of the proposed approach is evaluated by learning a maintenance policy for the beams structural category within a bridge in the Quebec province, Canada.