Abstract:

The optimal maintenance policy for deteriorating systems has been studied extensively. In almost literatures, the system deteriorates according to a stationary transition law. However, in real situations, systems can also deteriorate due to age. In such cases, the transition probabilities of the system should differ for different ages of the systems. For example, the rate of deterioration of a transformer increases with its age. Such systems are called "non-stationary deteriorating systems."

This talk will focus on the condition monitoring maintenance for an aging system of which the deterioration undergoes as a non-stationary Markov process. The optimal decision policy is investigated, and the structural properties of the resulting optimal expected cost function are obtained. These structural properties establish the existence of an optimal control limit policy with respect to both the system's deterioration and age under some assumptions. Furthermore, the monotonic property of control limits is also obtained.

If the optimal decision policy can be limited into the set of control limit policies, the tremendous amount of calculation time required to find the optimal decision policy would be reduced. Furthermore, the monotonicity of control limits can reduce the computational efforts substantially by simplifying the algorithm and reducing the computation errors.