Selection of Prior Distributions for Bayesian Reliability Assessment
Abstract
A necessary part of the Bayesian method is the choice of the prior distribution. In reliability assessment it is often desirable to obtain prior distributions for the components of a system which give a uniform prior distribution for the whole system when a moment-matching technique is used. There are an infinite number of choices, but it can be proved that component prior distributions can always be chosen to fall within a certain range, and still give the desired system outcome provided the desired system prior distribution is well behaved. Results are obtained by considering the case for a system composed of two independent components in series, a system composed of two independent components in parallel, and the more general system composed of a combination of the preceding two. (Modified author abstract)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1973
- Accession Number
- AD0762525
Entities
People
- Don R. Halverson