| Summary: | 碩士 === 國防管理學院 === 國防資訊研究所 === 90 === Traditional diagnosis methods mostly depend on people’s knowledge and experience, and that makes maintenance affected by people. Besides, it will increase the difficulty of designing a diagnosis system, because it’s tough to systemize people’s experience and professional knowledge.
In this thesis, we propose an ontology-based diagnosis architecture and provide more efficient and precise diagnosis information by use of logic inference. We describe how to design ontologies representing relevant knowledge about a domain, a task, and a problem-solving method. In addition, an ontology linker and an ontology adapter are used for connecting ontologies to reduce the interdependence among ontologies and to increase reusability. We also explore the relationship between symptoms and causes and finish a diagnosis task by setting weight and applying logic inference.
Furthermore, a learning mechanism is applied to accumulate previous diagnostic histories and adjust the weight to produce equipment-specific information and make a more precise diagnosis. By updating the knowledge base, the architecture can provide an appropriate diagnosis when getting new fault information about the equipment.
|
|---|
