Bayesian analysis for system reliability inferences

Due to declining defense budgets, operational test and evaluation programs for military systems have been down-one significant result is less system operational test data. This leads to lower classical confidence inferences on critical measures of performance such as system reliability. This paper explores the use of Bayesian methods to compensate for this decrease in test data by integrating developmental test and evaluation data into the operational test and evaluation phase. Specifically, the reliability of a series system, a parallel system and a classic bridge system are examined over a range of mission times. Components in this study are assumed to have exponential lifetimes. Using simulated data for the three systems, the Bayesian methods produce tighter probability intervals for the system reliability than the confidence intervals found through classical inference methods.