Reliability estimate of probabilistic-physics-of-failure degradation models

It is universally accepted that physics-based models provide more accurate estimates of reliability than statistics-based model especially when the physics of failure of the units are well-understood. However, the underlying assumptions of such models regarding the deterministic nature of its parameters limit their applications, implementations and generalization. In this paper, we propose a physic-based accelerated degradation model that considers probability distributions of its parameters. The degradation path is then determined, and the first passage distribution, failure time function and its parameters' estimation are obtained accordingly. The model is then used to estimate competing risks at normal operating conditions. A numerical example is provided to demonstrate the use of the model and validate its estimates.