Abstract:
|
Design of experiments (DOE) is a useful tool when planning tests of military systems. Complications arise when the test cannot be completed all at once and, rather, it needs to be performed in multiple phases. For these cases, sequential DOE could be used either by planning an open loop or close loop design. Bayesian techniques can be used when planning the latter. By doing so, prior information might be incorporated in the design of the first phase, and the information gained from a previous run or test phase could be incorporated in the design of the following phase. Consequently, the analyst could obtain as much information as possible from each run. This talk will present a case study example of a missile system where the test runs are limited and, therefore, a sequential Bayesian DOE is valuable.
|