Optimal Reallocation and Replacement Policy for Degradation Management with Unbalanced Workload

摘要

The units in industrial systems can be assigned to different workload, leading to unbalanced unit degradation levels in a system. For the sake of degradation management, we can change the user's behavior to balance the unit degradation levels. On the other hand, this strategy may be difficult or even impossible to implement in some cases. To address the issue, we integrate the unit reallocation into a condition-based maintenance (CBM) framework. Under the proposed reallocation/replacement policy, we perform a periodic inspection and the units are reallocated when their degradation levels differ too much from each other. We use a multivariate Wiener process to capture the dynamics of degradation, where the correlation between unit degradation is considered. The degradation with reallocation/replacement is formulated as a semi-regenerative process and we further show its embedded Markov chain is Harris recurrent. By computing the stationary distribution of degradation levels at inspection epochs, the system expected cost rate can be derived as a function of the reallocation and the replacement thresholds. Through the generalized pattern search algorithm, the optimal reallocation and replacement thresholds are obtained by minimizing the expected cost rate. We apply the proposed reallocation CBM policy to an automobile tire system and show that the reallocation can significantly reduce the expected cost rate with the existence of unbalanced workload.