Optimal inspection-based preventive maintenance policy for three-state mechanical components under competing failure modes

Abstract

The degradation of mechanical components is non-stationary and their degradation level can only be detected through offline discrete inspection in most cases. Aiming at the effect of the non-stationary feature and the delay of state detection caused by discrete inspection on long-run operation cost, this paper focuses on proposing an optimal inspection-based maintenance policy for three-state mechanical components subject to competing failure modes. A double-Wiener-process degradation model is established to describe the two operation states, which includes two Wiener process models under the same law but with different parameters. In addition, a preventive maintenance policy including a degradation threshold, an age threshold and a degradation control limit are adopted. A five-scenario probability-based model is proposed to describe the state evolution during one inspection interval and then the analytical model of the impact of delay on detecting state-transition and degradation-level on preventive maintenance policy is proposed. An optimization model of an inspection- based maintenance is established. Finally, sensitivity analysis in terms of the decision variables for state-transition time distribution and parameters in wear-out state is carried out based on numerical examples and the efficiency and superiority of the proposed policy is demonstrated by comparing with the current two-state maintenance policy.