变压器油中溶解气体故障诊断的改进模糊算法

Improved Fuzzy Algorithm for Fault Diagnosis on Dissolved Gas in Transformer Oil

变压器故障原因十分复杂,故障现象与故障机理间的联系存在着模糊性和不确定性。现有的经典故障诊断算法应用广泛,但存在一些固有缺点。为此,引入遗传算法,对模糊迭代自组织数据分析算法(iterative selforganizing data analysis techniques algorithm,ISODATA)进行改进,提出优化诊断方法,提高了算法效率,降低了ISODATA对于初始聚类的依赖性。结合具体诊断案例,对改进前后诊断方案各方面指标进行比较分析,证明所提方案的准确性、高效性。采用遗传算法改进的模糊ISODATA更符合实际需要,可方便地应用于对油浸式变压器的故障诊断。

Reasons for transformer faults are very complex and there are fuzziness and uncertainty in connection between fault phenomenon and mechanism.Existing classical fault diagnosis algorithm is widely applied but there are some inherent demerits.Therefore,genetic algorithm was introduced for improving iterative self-organizing data analysis techniques algorithm（ISODATA）and optimizing diagnosis method was proposed as well which could improve effectiveness of the algorithm and reduce dependency of ISODATA on initial cluster.Combining specific diagnosis cases,various index of the diagnosis schemes before and after improvement were compared and analyzed that verified veracity and high efficiency of the proposed scheme.It was proved that the fuzzy ISODATA based on genetic algorithm was more in accordance with practical requirements and convenient to carry out fault diagnosis on oil immersed transformer.