基于QPSO-SVM与DGA五边形解释工具的变压器故障诊断方法

Fault Diagnosis Method of Transformer Based on QPSO-SVM and DGA Pentagonal Interpretation Tool

鉴于变压器油中溶解气体分析(dissolved gas analysis,DGA)五边形解释工具存在依赖现场经验、准确率较低和分类边界过于绝对化等问题,提出了基于量子行为粒子群优化支持向量机(quantum-behaved particle swarm optimization support vector machine QPSO-SVM)与DGA五边形解释工具的变压器故障诊断方法。首先,基于变压器油中溶解气体数据,计算了Duval Pentagon1特征气体相对百分比的质心坐标和Mansour Pentagon特征气体相对百分比的质心坐标。其次,构建了QPSO-SVM-Duval Pentagon 1和QPSO-SVMMansour Pentagon变压器故障诊断模型。最后,对不同变压器故障诊断方法进行对比分析。仿真结果表明,QPSO-SVM-Duval Pentagon 1和QPSO-SVM-Mansour Pentagon变压器故障故障诊断方法准确率高于95.00%;QPSO-SVM-Duval Pentagon 1与QPSO-SVM-Mansour Pentagon相比准确率高、计算复杂;所提出方法与常规的DGA五边形解释工具和传统QPSO-SVM变压器故障诊断方法相比,变压器故障诊断准确率更高。

In view of such problems of the transformer oil dissolved gas analysis(DGA)pentagonal interpretation tool as relying on field experience,low accuracy and too absolutist classification boundary,the fault diagnosis method of the transformer based on quantum-behaved particle swarm optimization support vector machine(QPSO-SVM)and DGA pentagonal interpretation tool is proposed. First,the centroid coordinates of the relative percentage of Duval Pentagon1 characteristic gas and the centroid coordinate of the Mansour Pentagon characteristic gas are calculated based on the data of dissolved gases in transformer oil. Then,QPSO-SVM-Duval Pentagon 1 and QPSO-SVM-Mansour Pentagon fault diagnosis models of transformer are constructed. Finally,the fault diagnosis methods for different transformers are compared and analyzed. The simulation results show that the accuracy of QPSO-SVM-Duval Pentagon 1 and QPSO-SVM-Mansour Pentagon fault diagnosis method for transformer is higher than 95.00%. Compared with QPSO-SVM-Duval Pentagon 1,QPSO-SVM-Mansour Pentagon 1 has high accuracy and complex calculation. And compared with the conventional DGA pentagonal interpretation tool and the traditional QPSO-SVM fault diagnosis method for transformer,the proposed method is more accurate.