Model-based diagnosis for continuous process supervision: The Alexip experience

The Alexip knowledge-based system uses model-based reasoning to analyse Alphabutol petrochemical units' dynamic behaviour. A good understanding of what is going on is needed to suggest the corrective actions which must be taken to maintain the unit in a desirable state. This means, as a number of variables are involved, that the system must be able to take into account factors such as time-delays, combination of upsets, noisy data. In this paper we discuss the diagnostic part of the knowledge-based system. The method we present involves a qualitative description of single events that may occur in the process unit and a general reasoning mechanism (written in first-order logic) that achieves diagnosis based on the qualitative description. Since the reasoning is totally general, the method is generic and can be applied to any process. Events will only need to be described according to the procedure we define.