Informative Bayesian Quantification of Design Reliability Based on Test Characteristics and Test Results
Volume 9, Number 6, November 2013 - Paper 09 - pp. 677-687
WILLIAM E. VESELYNASA Headquarters, Washington, DC 20545, U.S.A.
(Received on February 20, 2013, revised on March 13 and April 09, 2013)
A common practice is to use random sampling statistics to estimate the demonstrated reliability of a new system from the number of successes and failures. This can grossly overestimate or underestimate the true reliability because of the naïve assumptions involved. A more detailed approach is presented which accounts for the characteristics of the test, the faults identified and the corrections made. The Bayesian framework accommodates a prior reliability estimate and comprehensively handles uncertainties. The examples show the wide areas of application from determining test strategies to tracking reliability growth. The applications described are believed to be new. The approach needs to be more widely implemented because of its features.
Click here to download the paper.
Please note : You will need Adobe Acrobat viewer to view the full articles.