zbMATH — the first resource for mathematics

Some new results on the confidence limits for reliability parameters in the case of no failures. (English) Zbl 0965.62086
Summary: In reliability tests, Type I censored data are common. For products of high quality, real life length is much longer than the censoring time. A group of items of some kind of product are put into reliability experiment. In case every item is censored, there is no failure. We refer to such data as No Failures. We construct lower confidence limits for life characteristics (such as average life, reliability, reliable span). This is done under the assumption of Weibull distribution, lognormal distribution and nonparametric distribution family in the case of no failures.
62N05 Reliability and life testing
62N01 Censored data models
62F25 Parametric tolerance and confidence regions