基于NRS的GWO-SVM变压器故障诊断方法研究

Research on GWO-SVM Transformer Fault Diagnosis Method Based on NRS

针对油中溶解气体分析法(DGA)不能有效反映变压器的不同故障且诊断准确率低的问题,通过邻域粗糙集(NRS)对变压器故障数据比值进行约简,得出一组新比值作为诊断样本,进而利用灰狼算法(GWO)与支持向量机(SVM)结合的模型进行故障诊断。实验分析表明,利用NRS对变压器故障数据约简能够有效提高变压器故障准确率,同时验证了GWO-SVM模型对于变压器故障诊断的良好适用性。

Dissolved gas analysis(DGA)in oil cannot reflect the different faults of the transformer effectively,and the diagnosis accuracy is low.This paper simplified the ratio of transformer fault data by using neighborhood rough set(NRS)to solve this problem.It derived a new set of ratios as a diagnostic sample from the simplified data.Furthermore,it used the gray wolf optimize(GWO)combined with the support vector machine(SVM)model for fault diagnosis.The experiment analysis shows that the use of NRS to simplify transformer fault data could effectively improve the accuracy of transformer faults.At the same time,it verifies the applicability of the GWO-SVM model for transformer fault diagnosis.