Importance Measures for Highly Reliable Systems.
Abstract
It is often desirable to maximize the reliability of a coherent system subject to resource constraints. However, for complex systems a formal optimization may not be feasible. For such systems, an approach which is often taken is to focus the resources on some small subset of components which are hopefully most critical to the system reliability. Such considerations lead to the notion of measures of component importance (re system reliability). These measures may take the form of relative rankings among components or may be absolute measures. Both mathematical and engineering definitions of component importance exist. The mathematical definitions typically have the feature of having desirable mathematical properties. The engineering definitions typically have the advantage of being easily calculated and of requiring a minimum of information. In this paper some mathematical and some engineering measures of component importance are compared. Some results are established which give some guidelines as to when one measure is preferable to the others. These results also make more explicit the underlying assumptions of the various measures.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 27, 1975
- Accession Number
- ADA016599
Entities
People
- David A. Butler
Organizations
- Stanford University