Engineering applications of Bayesian statistical methods
This dissertation makes Bayesian contributions to engineering statistics in three basic areas. These are methods for combining information, modeling repairable system reliability, and designing experiments.;A recursive Bayesian hierarchical model (RBHM) is presented. An RBHM can be used to combine information from physical data, data from a computer model of a process, and experts. In an example involving a fluidized bed process, an RBHM is used to estimate location and scale biases of one source of information for another.;The need to document the reliability of the Blue Mountain supercomputer motivates the work on system reliability. A detailed reliability analysis of this supercomputer is presented, using a Bayesian hierarchical nonhomogeneous Poisson process model. Further, some flexible new families of intensities for nonhomogeneous Poisson processes are defined and Bayes inference for them is discussed.;Finally, the problem of estimating expected information gain for planned data collection is considered. Two methods of estimation are applied to the so called random fatigue-limit model, a 5 parameter model important in some materials engineering applications.