Temporal Reasoning in Model-Based Diagnosis

摘要

Recently, research on model-based diagnosis of technical systems has beendirected to those systems, in which time doesn't play a role, or in which this factor is not taken into account. For these static systems, many methods have been developed which identify diagnoses in a rather efficient way. Diagnosing dynamic systems leads to several problems. First of all, because of the dynamic behavior of these systems, a mechanism for reasoning with time is needed. Another problem is feedback in dynamic systems. In such systems, model-based diagnosis has a low discriminating power. In this report, a diagnostic method is described for diagnosing discrete dynamic systems. The basic idea for this method is symbolic simulation, like it is used for diagnosing static systems. Temporal reasoning will be used partially. This method seems to be able to diagnose discrete dynamic systems rather well. When diagnosing systems with feedback, the problem concerning the low discriminating power is still prevalent. For this problem, a possible improvement is presented, but future research will still be necessary.