Inference and dynamic decision-making for deteriorating systems with probabilistic dependencies through Bayesian networks and deep reinforcement learning

Pablo G. Morato, C. P. Andriotis, K. G. Papakonstantinou, Philippe Rigo

Abstract: Highlights•An efficient modeling framework for inference and decision-making is proposed.•Environments under general deterioration correlation are effectively treated.•Factored POMDP and DRL methods are integrated into a unified algorithmic framework.•Underlying system dependencies are intrinsically considered by POMDP-DRL policies.•POMDP-DRL policies offer substantially lower costs compared to their counterparts.