DocumentCode :
894040
Title :
Maximum Entropy and Reliability Distributions
Author :
Teitler, S. ; Rajagopal, A.K. ; Ngai, K.L.
Author_Institution :
Naval Research Laboratory, Washington DC
Volume :
35
Issue :
4
fYear :
1986
Firstpage :
391
Lastpage :
395
Abstract :
An effort is made to show the relevance and usefulness of the principle of maximum entropy to reliability considerations. The constraints entering into the maximum entropy principle are identified as a class of sufficient statistics which determine the unknown parameters in the probability densities that occur in the most commonly used reliability models. In this way, the maximum entropy principle is shown to be completely compatible with prevailing practice in failure analysis. It is also pointed out that the differential entropy is equal to unity minus the expectation of the natural logarithm of the hazard rate. Maximization of the differential entropy is therefore equivalent to minimization of the expectation of the logarithm of the hazard rate. Behavior of the differential entropy under transformation of variable is used as an indicator of change or lack of change of conditions of failure.
Keywords :
Entropy; Failure analysis; Hazards; Laboratories; Probability distribution; Random variables; Reliability theory; Statistical analysis; Statistical distributions; Statistics;
fLanguage :
English
Journal_Title :
Reliability, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9529
Type :
jour
DOI :
10.1109/TR.1986.4335479
Filename :
4335479
Link To Document :
بازگشت