Stochastic Accelerated Degradation Models Based on a Generalized Cumulative Damage Approach

Abstract

A general methodology for stochastic degradation models is introduced that allows for both hard and soft failures to be taken into account when conducting parametric inference on lifetimes. Due to the development of engineering and science technology, modern products have longer lifetimes and greater reliability than ever before. Thus, it often takes more time to observe failures under normal-use conditions. Accelerated tests have been developed in order to deal with this lifetime-to-failure increase. Accelerated tests decrease the strength or lifetime to failure by exposing the specimens or products to harsh conditions. This exposure results in earlier breakdowns. Modelling these accelerated tests requires the use of stochastic degradation models with accelerating explanatory variables. By using a generalized cumulative damage approach with a stochastic process describing degradation, we develop stochastic accelerated degradation models which handle failure data consisting of both hard and soft failures.