基于D-S证据理论的变压器故障诊断

针对电力变压器故障征兆与原因之间错综复杂的关系,以及单一变压器故障诊断算法精度有限的问题,本文提出一种在D-S证据理论的基础上,结合灰关联熵法和加权K邻近算法的变压器故障诊断新方法。该算法以油中溶解气体分析方法(Dissolved Gases Analysis,简称DGA)为基础,通过灰关联熵法和加权K邻近算法构建证据理论的基本可信度赋值函数,然后利用证据组合规则产生更为可靠的证据信息;最后根据基本可信数最大值确定变压器故障类型。变压器故障诊断实例结果表明该算法能够准确判断出变压器的故障类型,证明了该算法的可行性和有效性。

Fault Diagnosis of Transformer Based on D-S Evidence Theory

Because the relationship between fault symptom and failure cause of power transformer is complex and the fault of transformer is not always diagnosed accurately by using single method,a new fault diagnosis method by combining grey association entropy method and weighted K-NN algorithm is proposed based on the D-S Evidence Theory in this paper.On the basis of the dissolved gases analysis(DGA),the basic credit assignment function of evidence theory is built by grey association entropy algorithm and weighted K-NN algorithm.Then,more reliable evidence information is generated by using of evidence combination rule.In the end,the fault of transformer is diagnosed according to the maximum basic credit value.The example of fault diagnosis of transformer testifies the feasibility and effectiveness of proposed algorithm,which can accurately diagnose the transformer fault.