融合集对分析和关联规则的变压器故障诊断方法

A Fault Diagnosis Method of Power Transformers by Integrated Set Pair Analysis and Association Rules

结合集对分析和关联规则各自的特点,提出了一种电力变压器故障诊断的新方法。该方法通过分析变压器运行中各故障征兆参数之间的关联性,建立了故障类型集合;通过对比计算关联规则的支持度与置信度,同时引入变权公式,得到故障类型和故障征兆的权重系数,有效避免专家意见或经验的主观性的问题;根据集对分析的可扩展性,采用5元联系度提高变压器故障诊断中处理不确定性因素的精度。实例对比分析表明,与关联规则、集对分析方法相比,该方法具有较好的诊断效果,在多故障问题处理中也表现优异。

A new power transformer fault diagnosis method based on set pair analysis（SPA） and association rules was presented. Via analyzing the associativity of fault symptoms, the set of fault types could be built up. Via computing the support degree and confidence degree of the association rules, and introducing variable weight formula, the constant weight coefficients of fault symptoms were got. The subjective problem of the expertise or experience was effectively avoided. According to the scalability of set pair analysis, this paper used 5-element connection degree. It could improve the accuracy of uncertainty factors in power transformer fault diagnosis. The analysis results of an example transformer showed that the presented method had to SPA and association rules and it excelled in multi fault diagnosis.