Adaptation-Guided Retrieval for a Diagnostic and Repair Help System Dedicated to a Pallets Transfer

In this paper, we describe a CBR approach for failure diagnosis of a pallets transfer. Adaptation phase is the key problem of the case-based reasoning system conception. This paper is a contribution to fill this gap in the equipments diagnostic and repair help. Retrieval step guided by adaptation is proposed, as a result similarity measures associated with an adaptation measure are proposed. These two measures will make it possible to select among the retrieved cases the most adaptable case. Then, an adaptation algorithm is proposed and will rely on a descriptors hierarchy, a context model as well as the dependences between problem and solution of the source cases. A feasibility study of the proposed algorithm is made on a real industrial diagnosis case. Three scenarios are treated in this study concerning various dependency relation values and belonging to the hierarchical classes of descriptors.