The path of a commercial aircraft engine through the flight envelope is generally quite predictable, and the bulk of operational time is at steady state; conversely, military engines are frequently used in dramatically different ways from flight to flight, with very little - if any - time spent at steady state. Thus, characterizing the health of military engines is much more difficult, and requires a very different approach than the one used for commercial engines. This paper describes an artificial intelligence approach for identifying semi-steady state points in a military engine, and for using the data collected at those points for characterizing engine health.