EVALUATING THE CONFIDENCE INTERVAL FOR THE RELIABILITY OF GENERAL MULTICOMPONENT SYSTEMS,
Abstract
The combining of components (e.g., diodes, transistors, semiconductors, etc.) into a complex system has presented the problem of estimating the reliability of the entire system with some specified degree of confidence. Various exact and approximate methods exist for very special cases. This paper, however, reviews some of the current simulation procedures and presents a method for solving the more difficult general problem. Three forms of data input are considered. (1) Point estimates of the parameters of the assumed underlying distributions for the component failures may be made from sample data (examples are given for the binomial and exponential distributions). (2) The lower or upper reliability limit (with appropriate confidence level) may be specified for each component with no information about the sample results. (3) A subjective assessment of a prior distribution of component failure times may be given along with (1) above. A Monte Carlo simulation technique, can take any combination of these three types of data input and compute an estimate of the reliability of a simple or complex system with a relatively small error. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1970
- Accession Number
- AD0709430
Entities
People
- Larry J. Massa