| Summary: | 碩士 === 國立中央大學 === 統計研究所 === 101 === Accelerated life testing (ALT) is a process of testing products by subjecting it to strict
conditions, in order to observe more failure data in a short time period. In this thesis,
we consider the ALT of series system, each consists of two components whose life time
distributions follow independent exponential distributions. Optimal designs on the sample
allocation for the two-level constant-stress ALT(CSALT) and on the time for changing stress
levels for the two-level step-stress ALT(SSALT) are considered based on V-optimality, D-
optimality and A-optimality, respectively. Under Type-I censoring, it shows, by numerical
results, that the optimal SSALT is better than the optimal CSALT in terms of the resulting
objective functions. We also prove that the two optimal ALTs are indeed equivalent without
censoring. In addition, we use the optimal CSALT as the baseline ALT to obtain an equivalent
SSALT plan. A real data is analyzed to demonstrate the performance of both ALT plans
under the three optimality criteria as well as the equivalent test plans.
|
|---|
