Abstract:
|
Proper experimental design is a crucial element of product reliability assessment. Optimized reliability performance and faster test completion can be achieved by different forms of Accelerated life Testing (ALT) under different experimental settings. In ALT, experiments usually start with relatively high stress levels. However, too high levels may introduce units to sudden shocks and cause mass failures. In this work, we perform a comparison study between adaptive step-stress ALT (ada-SSALT) and simple step-stress ALT (SSALT) models. A set of parameters are tuned numerically to reduce the estimation bias and to improve the precision along with their confidence intervals based on the approximate and empirical approaches. This is discussed under several design criteria including D, C, A, M, and E. This study assumes that the units’ lifetime follows an exponential distribution, and a log-linear relationship exists between the standardized stress level and the mean unit time to failure (MTTF). The computational results suggest superiority of ada-SSALT. A real case study shows that the proposed algorithm outperforms the simple SSALT in the presence of a shock effect.
|