Abstract: Adaptive scheduling selects scheduling rule in a real-time manner according to system states. Scheduling knowledge, which is ordinary expressed as if-then format and performs mapping from system states to scheduling rules, plays a central role in this scheduling paradigm. A concept of stability for scheduling knowledge is presented, which is defined as performance sensitivity of the scheduling system to variety of manufacturing tasks. A necessary condition that a stable scheduling knowledge should satisfy is proposed, and then is used to guide selection of scheduling knowledge.Simulation experiments demonstrate that a better adaptive performance can be obtained if the scheduling knowledge is selected according to the stability criterion.

Key words: FMS Adaptive scheduling, Knowledge acquirement Stability