3. Discussion

In this article, we present a number of techniques for finding diagnoses using artificial intelligence approaches. Of course, the list of techniques presented here is not exhaustive.

The table at 5 summarizes these techniques, describing the knowledge used, the advantages and disadvantages. It should be noted that no approach is perfect. For example, model-based approaches enable diagnostics to be performed on the basis of a description of a device's correct operation. This description is, in theory, available from the device designers. The undeniable advantage of this approach is that it can be applied without waiting for faults to occur, which is necessary to build up diagnostic expertise. However, this approach relies on the accuracy of the model used. If it contains errors, the diagnoses obtained...