Causal-Based Diagnosis of a Hydraulic Looper

Abstract This paper deals with a particular diagnostic agent (DA) within the MAGIC system (Multi-agent architecture for diagnosis and operation support in complex installation), namely the causal knowledge based diagnostic agent (KBDA). The MAGIC system requires the symptoms elaborated by DAs to be sent to the Diagnostic Decision Agent (DDA) that performs the fault isolation. The global architecture of a DA is first presented. Then, a general view of the detection algorithms that have been implemented in the KBDA is given. Beside the detection test result itself and the detection time, two other indicators, the alarm decision membership function and the model validity, are part of the symptom and will be presented here. This DA is used to diagnose a part of a hot rolling mill, namely the hydraulic looper in a finishing mill.