An Age-State-Dependent Degradation Model

Degradation modeling plays a key role in accelerated degradation test data analysis and condition-based maintenance. The degradation rate of a degradation process usually depends on both unit's age and its state. Several agestate-dependent degradation models have been proposed in the literature. A main drawback of these models is their intractability. In this paper, we propose an analytically tractable age-state-dependent degradation model. The proposed model is defined in terms of degradation rate, whose mean function is represented by a bivariate power-law model. The degradation increment is modeled by a non-homogeneous gamma process, whose mean function depends on both age and degradation level and has a closed-form expression. The model includes age-and state-dependent models as its special cases. An approach is proposed to make selection among the general case and special cases based on the Akaike information criterion. A real-world example is used to illustrate the flexibility and usefulness of the proposed model and approach.