Abstract:

According to the concept of General Diagnostic Engine (GDE), an approach to identify minimal conflict sets is proposed. It is based on the Assumptionbased Trust Maintenance System (ATMS). By defining negativenode and new rules, the incompleteness of ATMS is overcome. The simplified entropy diagnostic measurement is realized using crude probability estimation; and a method based on simplified entropy using dynamic ε policy is put forward. Furthermore, details of dynamic update of candidate diagnoses after each measurement are given. This approach is incremental through propagating environments. It possesses the characteristics of correctness, completeness, consistency, all solution minimization and low diagnostic cost. Simulations also show that this approach is effective, general and easy to be implemented and expanded as well.

Key words: model based diagnosis, assumption based truth maintenance system, minimal conflict sets, measurement;dynamic &epsilon, policy