Abstract
It is an important concern for industries on how to avoid negative effects due to machine deteriorations and failures. Maintenance executed preventively is such a choice to maintain machine in proper operations. In nowadays technology, machine can be real-time monitored by equipping on-line inspecting equipments. Such equipments can be connected to a computer center which can help to determine the optimal action at each inspection time through a suitably designed algorithm. Hence, dynamic (real-time, time and state dependent) preventive maintenance policies can be realized or carried out.The requirements for such a dynamic preventive maintenance policy are expressed as follows;(1)Choose a suitable healthy index H. Such index can be calculated in terms of process parameters which can be obtained by inspection equipments.(2)The system state can be determined by a suitably chosen(3)The system can be modeled by a non-homogeneous Markov chain.(4)Upon each state, several preventive actions can be chosen.(5)The maintenance time is taken into consideration.(6)The inspection is ignored.(7)According to data collected, evaluates the transition probability matrices.(8)The optimum maintenance policy is to choose at each inspection time the action which will have minimal expected total cost during a long time interval in the future.