Application of Bayesian Techniques to Reliability Demonstration, Estimation and Updating of the Prior Distribution.
Rept. for Jun-Aug 78,
ROME AIR DEVELOPMENT CENTER GRIFFISS AFB NY
Pagination or Media Count:
A method is presented for the estimation of the shape and scale parameters of an inverted gamma prior distribution of the mean-time-to-failure for equipment having exponential time-to-failure distribution. This method, akin to the Maximum Likelihood Method, allows the use of all sorts of existing failure data on the equipment in question, provided a certain sufficient condition is satisfied. Further, this method we call it the Generalized Maximum Likelihood Method is usable to update the prior distribution, when new failure data become available. In the long run, this updating process will give rise to a solid prior, which can confidently be used in Reliability Demonstration. Various facets of the sufficient condition for the applicability of this estimation method are exposed, the variance--covariance matrix of the estimators is given under various randomness assumptions and some numerical considerations are discussed. There is a brief discussion of alternate estimators in the case of a truncated test data. Author
- Statistics and Probability
- Manufacturing and Industrial Engineering and Control of Production Systems