Community Research and Development Information Service - CORDIS


The estimation of failure rates of heterogeneous Poisson components from data on times operated to failure is reviewed. Particular emphasis is given to lack of knowledge on the form of the mixing distribution or population variability curve. A new nonparametric empirical Bayes estimation is proposed which generalises the estimator of Robbins for different times of observation for the components. The behaviour of the estimator is discussed by reference to two samples typically drawn from the CEDB, a component event database designed and operated by the JRC Ispra.

Additional information

Authors: ARSENIS S P, JRC Ispra (IT)
Bibliographic Reference: Paper presented: International workshop series on advanced topics in reliability and risk analysis, Annapolis (US), October 20-22, 1993
Availability: Available from (1) as Paper EN 37950 ORA
Record Number: 199510540 / Last updated on: 1995-04-21
Original language: en
Available languages: en