In this study, the degradation path of a highly reliable product is modeled by Gamma process. The generalized Eyring model is used to link the shape parameter of Gamma process with the accelerated variable over experimental runs. Sensitivity of the sample size and measurement frequency on the parameter estimation of single variable constant-stress accelerated degradation test model is investigated through Monte Carlo simulations. Simulation schemes are set to accommodate a realistic accelerated degradation test experiment of high power light emitting diodes. A planning strategy for the smallest sample size and measurement frequency of the accelerated degradation test is reached with smaller bias and mean squared error of maximum-likelihood estimates of model parameters.
Relation:
ICIC Express Letters, Part B: Applications 6(3), pp. 737-742