Bayesian Design of a D-Optimal Accelerated Degradation Test Considering Random Effects

Accelerated degradation testing (ADT) is a viable option for organizations to quickly evaluate the reliability of highly reliable products. To improve the precision of reliability estimation by incorporating prior knowledge about the products, Bayesian ADT plans have been studied for simple stochastic degradation processes. Such plans are often more robust than those based on maximum likelihood estimation. However, the unit-to-unit variation of a newly developed product raises a big challenge in planning ADT. In this paper, we study the Bayesian design of optimal ADT plan considering random effects in the parameters of a monotonic degradation process. Specially, the proposed D-optimal test plan is optimized by determining the number of test units allocated at each stress level, and the number of degradation measurements and the censoring time for each test unit. A numerical example is provided to illustrate the performance of the proposed method in handling random effects and computational challenges.