Knowledge-Based Supervision is an outstanding Artificial Intelligence field contributing successfully to the progress of the control and supervion areas. Three essential factors characterize the function of a supervisor system: time constraints demanded from the supervision process, temporal updating of information coming from the dynamic system and generation of qualitative knowledge about the dynamic system.

In this work, an evolutionary data structure model conceived to generate, store and update qualitative information from raw data coming from a dynamic system is presented. This model is based on the concept of abstraction, in such a way an abstraction mechanism to generate qualitative knowledge about the dynamic system which the Knowledge-Based Supervisor is based on, is triggered according to some pre established considerations, among which real time constraints play a special role.