FCC Process Diagnosis Based on a Causal and Heuristic Approach

Abstract This paper presents a diagnostic system which gave encouraging results on real data scenarios and is tested on line on a FCC 1 pilot plant. This system uses two complementary approaches. A model based diagnosis is performed in order to detect and to localise faults on measured variables. Causal reasoning and fuzzy sets are used to localise faults. Then the operator knowledge is exploited both to suspect a fault on a physical component and to propose counteractions to perform.