| Summary: | 碩士 === 正修科技大學 === 電機工程研究所 === 99 === From the view of preventive measures, the work on earlier detection of the incipient fault of the fundamental equipment in the power system, especially the power transformers, has attracted quite much attention. As a vital device in the power system, the fault of the transformer will lead itself to a very wide range of outage of the system and maybe cause very serious damages to its surroundings due to its poisonous and combustible insulation oil inside. To understand operating conditions of power transformers, on-line monitoring and periodic examinations of the transformer must be rigorously performed to find the incipient fault inside and to prevent it from further deterioration as early as possible. Periodic maintenance of large power transformers includes chromatographic analysis of the insulation oil to measure the concentrations of diverse dissolved gases. According to the concentrations of the dissolved gases and other conditions of the transformer, the transformer fault diagnosis is conducted through dissolved gas analysis (DGA) approach.
To solve the fault diagnosis of power transformers, this thesis presents a data mining tool -- rough set theory based approach to extract fault diagnosis rules from diagnosis database of power transformers. The proposed rough set approach can directly describes the discovered knowledge by the extracted rules. The test results have shown that the proposed approach can extract simple and effective diagnosis rules from database, and the feasibility of applying the method to diagnose the incipient faults of power transformers has been demonstrated.
|
|---|
