Abstract:

The machinery condition monitoring community continues its search for the ``Holy Grail'' of prognostics. The problem is that there is no common agreement on two questions: ``What does the term prognostics mean in this context''? and ``What must be done to achieve it''? This paper differentiates ``predictive maintenance'' (in its generally accepted form) from ``condition-based maintenance'' (which requires prognostics). A justification for development and implementation of the more advanced and cost-effective condition-based maintenance is presented. A working definition of prognostics for the condition-based maintenance community is proposed. Further, a methodology for setting acceptable false alarm rates for both diagnostics and prognostics is presented to frame the problem for the research community in terms of the application environment, particularly for mission critical equipment. [See NOISE-CON Proceedings for full paper.]