Reliability estimation using doubly-censored field data

A frequently encountered problem in reliability verification is described, and several approaches to managing this problem are presented. The general problem is the desire to use field reliability data rather than laboratory data to evaluate equipment performance despite the deficiencies of reported field data. This problem is most severe when the equipment age is unknown at the start of a period of equipment monitoring and the tracking of the equipment is stopped after a relatively short observation interval. The use of the resulting data set to construct life distribution parameter estimate is described here. Several heuristic parameter estimation algorithms are defined. The set of proposed methods is reduced on the basis of tractability concerns, and the remaining method are applied to simulated data sets. For comparison, the corresponding unabbreviated data sets are analyzed using established methods. The comparison are made for life data that is negative exponential and for data that is Weibull for both increasing and decreasing hazard cases.<
>