A generalized three-type lifetime probabilistic models-based hybrid maintenance policy with a practical switcher for time-based preventive maintenance and condition-based maintenance

This paper proposes a generalized hybrid maintenance policy for maintenance scheduling with the help of both time-based maintenance and condition-based maintenance techniques, in which three-type product lifetime probabilistic models are utilized. A dispersion of lifetime estimates-based switcher is proposed to recommend the choice of time-based maintenance or condition-based predictive maintenance schedules in a real-time manner. Within the condition-based predictive maintenance policy, which is a part of the hybrid maintenance policy, a novel weighted average of the maintenance schedules is proposed to recommend maintenance acts, which are estimated based on two types of product lifetime probabilistic models namely type-II and type-III lifetime probabilistic models. The hybrid maintenance policy includes the classical time-based maintenance policy, the traditional condition-based predictive maintenance policy, and the proposed condition-based predictive maintenance policy as special cases. An extensive numerical investigation for the stochastic linear degradation model verifies the effectiveness of the proposed hybrid maintenance policy, highlighting the existence of a special space among time-based maintenance and condition-based predictive maintenance polices, which provides even better maintenance performance than a solely condition-based predictive maintenance policy.