Abstract:
The optimal plans are investigated for constant-stress accelerated life tests (ALT) under Type I progressive interval censoring when lifetimes are exponentially distributed. In particular, the number of removals at each inspection stage is random and assumed to follow a discrete uniform distribution. The maximum likelihood function and the associated Fisher information matrix are derived. The expression of the asymptotic variance of the estimated mean lifetime parameters is obtained. The optimal stress levels and allocation proportions which minimize the asymptotic variance of the estimators for different number of inspections are presented. Furthermore, sensitivity analysis is conducted through Monte-Carlo simulation to investigate the effect of mis-guessed lifetime parameters on the asymptotic variance.