Application of Bayesian Reliability Concepts to Cruise Missile Electronic Components.

摘要

The purpose of this thesis was to evaluate the applicability of Bayesian statistical methods to the problem of determining cruise missile component reliability. There were three objectives: 1) to develop models incorporating Bayesian reliability concepts that can be used to predict component reliability based on data available in a program transitioning from development to production; 2) to determine the model's validity in comparison with classical statistical models; and 3) to assess the accuracy of both approaches against actual cruise missile flight test history. A total of six models were developed for the failure rate of the Tomahawk Cruise Missile Guidance Set using both exponential and binomial distributions. The flight test data seemed to belong to another failure distribution, and was not useful as a measure of performance as had been proposed. The Bayesian Expert Information Model provided reasonable point estimates of the failure rate and markedly shorter 90% confidence intervals. In general, the Bayesian models had confidence intervals that were shorter than the classical statistical inference models, allowing a more accurate decision-making process. (jhd)