An Approach to Prognostic Decision Making in the Aerospace Domain

The field of Prognostic Health Management (PHM) has been undergoing rapid growth in recent years, with development of increasingly sophisticated techniques for diagnosing faults in system components and estimating fault progression trajectories. Research efforts on how to utilize prognostic health information (e.g. for extending the remaining useful life of the system, increasing safety, or maximizing operational effectiveness) are mostly in their early stages, however. This process of using prognostic information to determine a system’s actions or its configuration is beginning to be referred to as Prognostic Decision Making (PDM). In this paper we, first, propose a formulation of the PDM problem with the attributes of the aerospace domain in mind, outline some of the key requirements on PDM methods, and explore techniques that can be used as a foundation of PDM development. The problem of Pareto set viability, i.e. satisfaction of performance goals set for objective functions, is discussed next, followed by ideas for possible solutions. The ideas, termed Dynamic Constraint Redesign (DCR), have roots in the fields of Multidisciplinary Design Optimization and Game Theory. Prototype PDM and DCR algorithms are also described and results of their testing are presented.