Reinforcement learning in maintenance decision-making and optimization: a literature review

Abstract In today's dynamic industrial landscape, maintenance managers are increasingly seeking innovative approaches to enhance efficiency in maintenance decision-making and optimization processes. Among these approaches, Reinforcement Learning (RL) has emerged as a powerful tool, offering promising avenues for achieving optimal policies amidst the complexities of evolving environments and data intricacies. This paper presents a comprehensive literature review, aiming to illuminate the application of RL in the realm of maintenance optimization and decision-making support. Utilizing a systematic approach, relevant literature was carefully selected from leading databases including Scopus, Web of Science, and IEEE Xplore. Through rigorous analysis of the selected works, this study offers detailed insights into the current state of RL's deployment in maintenance optimization and decision-making, while also identifying prevailing limitations, challenges, and future trends in the field. By providing a comprehensive overview of RL's applications in maintenance optimization, this research contributes to a deeper understanding of its efficacy and potential, while also pinpointing key areas for further exploration and identifying major knowledge gaps. Ultimately, this endeavor seeks to catalyze advancements in maintenance strategies by addressing critical challenges and leveraging emerging trends in RL-based decision-making paradigms.