Ontology of fault

As mentioned above, model-based problem solving systems have own assumptions and limitations. In the case of diagnostic problems, each model-based diagnostic system detects a limited scope of possible faults. For example, GDE assumes that the topology of connections among components is not changed by the faults.

The goal of this research is to establish an ontology of faults, aiming at clarifying the scope of detectable faults of the diagnostic systems. The ontology defines several concepts such as "cause of faults". We believe there are two meanings of causes of faults, that is, the cognitive causes and the physical causes. The former specify locations of faults (i.e., component name) as results of cognitive interpretation based on hierarchical functional architecture. The later represent the states that physically cause the symptoms together with physical causal chains among them. We also define a set of classes of faults from the viewpoints of physical causes.

The fault ontology provides us the vocabulary for the followings;

• To specify the limited scope of detectable faults of the diagnostic systems (such as GDE).
• To propose such a diagnostic method that exhaustively generates fault hypotheses in wider range.
• The set of classes of faults in the ontology allows users to control the scope of suspected faults.

Major publications

1. An Ontological Analysis of Fault Process and Category of Faults, Proc. of Tenth International Workshop on Principles of Diagnosis (DX-99), June 8-11 1999, Scotland, Forthcoming
2. A Model-based Diagnosis with Fault Event Models, Proc. of Pacific Asian Conference on Expert Systems (PACES) 97, 1997