A DRL-based strategy optimization of usage and maintenance considering operational readiness

Operational readiness (OR) measures a system’s ability to respond to emergent missions and represents a key performance indicator. This paper proposes a joint optimization approach based on deep reinforcement learning (DRL) to determine optimal usage and maintenance strategies for a multi-unit multi-state system (multi-unit MSS) under OR requirements. During system operation, the planner must select units for subsequent usage and maintenance while accounting for periodic Spare part ordering. These decisions are conditioned on the current system state and are made at regular inspection intervals. The objective is to preserve the OR of the system while simultaneously meeting daily operational demands. This decision-making process is formulated as a Markov decision process (MDP). Within the DRL framework, a tailored method is propoesd to obtain the optimal joint strategy. A case study is conducted to validate the effectiveness of our method.