A monitored system consists of a core system and its monitor. A core system follows a general life time distribution, but its state, good or not good, cannot be observed directly. In order to infer the state of the core system, we use the information from the monitor whose states are stochastically related to the states of the core system. Using its information, we make the maintenance policy for the core system so as to minimize the total expected cost incurred by a sequence of actions. This paper shows the conditions under which an optimal maintenance policy for the above model has the form of a control limit policy. Also, a numerical example where the life time of the system follows a Weibull distubution is presented.
View full abstract