Portal de Periódicos da CAPES

A Fault Tree Generation Method using Qualitative Reasoning and its Application to a Protective Relay

1994; Institute of Electrical Engineers of Japan; Volume: 114; Issue: 4 Linguagem: Inglês

10.1541/ieejpes1990.114.4_410

1348-8147

Masanori Akiyoshi, Shogo Nishida, Masahiro Shiotani, Tadahiro Goda,

Model-Driven Software Engineering Techniques

Human experts diagnose artifacts by generating fault trees, i. e., finding possible faulty components and their cause-effect relations from observations. Lately such task seems to be much harder, as artifacts become complex. Therefore computational methods, such as model-based reasoning, are expected to cope with the complexity. This paper presents a method of generating fault trees that uses qualitative reasoning. The proposed method deals with “model generation” which relates to the structural complexity and “inference on symptoms” which relates to the behavioral complexity. First, a target model is generated by using a model library that includes component models and structural data that are originally drawn on a design information. Second, states of components under an abnormal symptom are inferred by using two procedures ; constraint propagation and state transition. Third, possible faulty components are detected by using inferred states and fault models. Finally, each possible faulty component is managed by using fault trees. The result of applying this method to a protective relay is also discussed.