A Generalized Confidence Limit for the Reliability of Parallel Systems Based on Type II Censored Data

摘要

Based on type Ⅱcensored data, using the concept of a generalized confidence interval due to Weerahandi, a generalized pivotal quantity for the reliability of a parallel system is obtained by the method of fiducially inference when survival time of the components follow the exponential distributions. Then, a generalized lower confidence limit is constructed for the reliability of a parallel system. The method is compared to the method of equivalent convention of the life data by simulation. Furthermore, a hybrid method is developed and the simulation study shows that the hybrid confidence limit is more efficient than two other methods.