Fuzzy qualitative modelling of pressure die-casting for fault diagnosis

We present an approach to modelling a die-casting machine for fault detection purposes. We used fuzzy qualitative reasoning to describe the behaviour of a part of the plant, because of the suitability of this technique for applications where parameter uncertainties are inevitable. Moreover, the advantage of using such a technique is that it gives a qualitative description of the plant, which mimics the way an expert operator would deal with it. A constraint-centred approach, based on the difference equations rather than differential equations was developed. The mechanism was then implemented in the Quintus Prolog language. It is still at an early stage of development, and refinements are to be carried out.