Reliability assessment model considering heterogeneous population in a multiple stresses accelerated test

Abstract

Heterogeneous population, a mixture of weak and strong subpopulations, is inevitable in some multiple stresses accelerated tests. Modeling the reliability of heterogeneous population in an accelerated test differs dramatically from that for a homogeneous setting. In this paper, a multiple stress reliability assessment model with heterogeneous populations is proposed, which includes verifying the presence of heterogeneous population, determining the number of subpopulations and separating populations based on Bayes classifier. The acceleration model structure is then specified, and the effects of different accelerating stresses are analyzed. A practical example is used to demonstrate the accuracy and flexibility of the proposed method. It is shown that reliability assessment without considering heterogeneity is heavily biased, and the sequences of stress sensitivity to different subpopulations are different. We also explain the phenomenon that the pseudo lifetime of smart electricity meter under some milder stress is shorter than that in harsher condition due to opposite effects on degradation characteristics among different stresses, and verify the phenomenon by the enhancement test.