RELIABILITY ESTIMATION UNDER PLAUSIBLE ASSUMPTIONS
Abstract
Most analyses of reliability problems assume that the form of the underlying failure distribution(s) is known; the parameter may be assumed known or unknown. Families of failure distributions commonly used are the exponential, gamma, normal, and lognormal. An expository survey is presented of recent research in reliability estimation based on assumptions made not simply for mathematical convenience but because they correspond to the physical situation. The class of statistical problems considered thus lies somewhere between parametric problems (in which the underlying failure distributions are assumed known up to a finite number of parameters) and nonparametric problems (in which no information is assumed available concerning the underlying failure distributions).
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1965
- Accession Number
- AD0619873
Entities
People
- Frank Proschan
Organizations
- University of California, Berkeley