Knowledge extraction for system design: application to a complex industrial scheduling problem

The article presents a cognitive approach for system design. This approach is developed at the STADE in collaboration with the LIASC. At the TrefilARBED industry, we treat a complex scheduling problem of wire drawing production. The aim of the industrial project is to design an automatic scheduling system. We use constraint logic programming (CLP) techniques to build this system (R.S. Bisdorff et al., 1995). These techniques have been already tested in complex industrial problems in previous projects at the CRP-CU (R.S. Bisdorff and S. Laurent, 1995). The design of this system is based on a cognitive study of a human agent´s task. The use of cognitive artefacts raises several questions. How can we simplify the realistic model? Which aspect has to be treated with the help of a cognitive artefact? What level of detail is needed for each artefact? Who is able to define these artefacts?