STATISTICAL INFERENCE FOR MULTIPLE STEP-STRESS MODEL WITH TYPE-II CENSORED DATA FROM THE GENERALIZED EXPONENTIAL DISTRIBUTION

This study introduces a Type-II censoring framework-based multi-phase stress accelerated life testing approach. We assume that the lifespan at the specified design stress follows a generalized exponential distribution. Furthermore, an inverse power law of the applied stress levels represents the scale parameter of the generalized exponential failure time under constant conditions. In order to estimate the model parameters using the cumulative exposure model's assumptions, the maximum likelihood method is employed. For these parameters, we build confidence intervals based on the asymptotic distributions of the maximum likelihood estimates. By applying the A-optimality criterion, we can determine the best time to introduce stress adjustments and the best way to implement censoring in accelerated life experiments. This paper concludes with numerical examples that show how the proposed methodology works.