Multiple-fault diagnosis under uncertain conditions by the quantification of qualitative relations

Among many fault diagnosis methodologies for chemical processes, the signed digraph (SDG) offers a simple and graphical representation of the causal relationship between process variables and has been widely used. The assumptions of SDG such as single-fault assumption and consistent path assumption, however, have limited its applications to real processes. This paper presents a multiple-fault diagnosis methodology utilizing many advantages of SDG. The diagnosis model is organized by carefully modifying SDG and searching the fault propagation path. The on-line diagnosis is performed through quantifying the qualitative relation of the fault propagation path, and this quantification enables the diagnosis to overcome the weakness of SDG under uncertainty and to diagnose multiple faults. Also, this work suggests how modified-CUSUM (cumulative sum) can be used to handle compensatory or inverse response. The proposed methodology is illustrated with a case study of a continuously stirred tank reactor (CSTR) with a heat exchanger.