General Framework for Learning Curve Reliability Growth Models

摘要

In reliability growth models, systems undergo an improvement in performance during prototype testing, as design changes are made, and operating procedures and environment are modified. In the learning-curve models, this improvement occurs continuously over time, and there is great interest in predicting the ultimate performance of the system, using only the epochs of the failures which occur early in the testing program. This paper constructs a general framework in which to analyze this problem, including as special cases many different model variations that have previously been analyzed. Numerical trails indicate the difficulty of using classical procedures to estimate ultimate performance; the maximum likelihood estimator is unstable for small testing intervals with a small number of systems on test, and is even inconsistent for a large number of systems. Bayesian procedures are recommended for implementation, as they can use the data from any testing protocol. (Author)