Adaptive Maintenance Policies for Aging Devices Using a Markov Decision Process

In competitive environments, most equipment are operated closer to or at their limits and as a result, equipment's maintenance schedules may be affected by system conditions. In this paper, we propose a Markov decision process (MDP) that allows better flexibility in conducting maintenance. The proposed MDP model is based on a state transition diagram where inspection and maintenance (I&M) delay times are explicitly incorporated. The model can be solved efficiently to determine adaptive maintenance policies. This formulation successfully combines the long term aging process with more frequently observed short term changes in equipment's condition. We demonstrate the applicability of the proposed model using I&M data of local transformers.