A Multi-AI Methods Based Model for Synthetic Diagnosis of Transformer Faults
Zhou Meng-ge · Dianli xitong zidonghua · 2005
Taking into account the efficiency of the rough set in incomplete information processing, the intuitiveness of querying and matching by case-based reasoning and the convenience of the ratio method, an approach based on multi-artificial intelligence methods for diagnosing transformer faults is proposed. By comprehensively considering the symptoms from DGA and electrical tests and analyzing large numbers of fault cases, a model for synthetic diagnosis based on the rough set, case-based reasoning and ratio method is developed. A three-layer structure has been constructed for this model to analyze the detailed fault of a transformer step by step. It is helpful to providing field personnel with effective advice even if the fault information may be incomplete. This model has such advantages as high speed of calculation, high correct rate, intuitional results, etc. Examples also show the effectiveness of the method.