论文部分内容阅读
目前变压器内部故障诊断诊断的有效性仍有待进一步研究。为此,提出了一种基于信息融合的多证据体变压器内部故障诊断新方法。该方法基于神经网络、SVM和S_Kohone聚类算法形成初级诊断结果出现分歧时的多证据体,判断证据体之间是否存在隐藏冲突,不存在隐藏冲突则优先采用证据分类折扣算法,否则采用证据主元的证据折扣算法对证据源进行修改,利用Dempster规则合成得出故障诊断结论。基于DGA和电气试验的实例验证表明采用的变压器内部故障诊断方法的有效性。提出的基于信息融合的变压器内部故障诊断新方法合理协调了证据体之间的信息冲突,有效融合了各种不同智能算法的判断结果,且故障识别正确率较融合前提高到88.65%。
At present, the effectiveness of the diagnosis of transformer internal fault diagnosis remains to be further studied. Therefore, a new method of internal fault diagnosis of multi-evidence transformer based on information fusion is proposed. This method is based on the neural network, SVM and S_Kohone clustering algorithm to form multi-evidence body when the primary diagnosis results disagree, to judge whether there is a hidden conflict between the body of evidence or not, if there is no hidden conflict, the evidence classification discount algorithm is prioritized, otherwise, Meta evidence discount algorithm to modify the source of evidence, the use of Dempster rules derived fault diagnosis conclusions. The example verification based on DGA and electrical test shows the effectiveness of the internal fault diagnosis method of transformer. The proposed new method based on information fusion in transformer internal fault diagnosis reasonably coordinated the information conflict between evidence bodies and effectively fused the judgment results of various intelligent algorithms. The accuracy of fault identification was improved to 88.65% before fusion.