Chapter 12 Robust inference under cyclic accelerated life tests ⍟ ⍟ This book has a companion website hosting complementary materials. Visit this URL to access it: https://data.mendeley.com/datasets/879xmdz3d8/1.

In the preceding chapters, we have presented different inferential methods developed for one-shot devices under constant stress levels. However, a cyclic stress test may be more practical in some situations, as the environment may be changing. In this chapter, we study the reliability of one-shot devices with Birnbaum–Saunders (BS) fatigue life under accelerated life tests (ALTs) with various cyclic temperature fluctuations by assuming different cyclic models. We introduce a family of robust estimators based on the density power divergence (DPD) measure as a generalization of the maximum likelihood estimator (MLE). Its robustness is demonstrated through an extensive Monte Carlo simulation study. We conclude the chapter with a numerical example on CSP solder joints, illustrating the practical utility of the proposed methods. We also provide a robust model selection criterion for the choice of the best cyclic model for a given data set.