AN INTEGRATED APPROACH FOR FAULT-DIAGNOSIS WITH LEARNING

This paper presents an integrated approach for diagnostic reasoning th at combines the strengths of both shallow reasoning and deep reasoning , and incorporates two different forms of learning into the diagnostic system. This new diagnostic approach should be widely applicable to a range of complex, computer-controlled systems. A case study is presen ted for a pumping system in a chemical process plant, in which several fault situations are simulated and correctly diagnosed by the diagnos tic system and the system learning capability is demonstrated.