基于深度自编码网络的电力变压器故障诊断

Transformer fault diagnosis based on deep auto-encoder network

基于深度自编码网络(DAEN),构建了分类深度自编码网络(CDAEN)模型。结合电力变压器在线监测油中溶解气体分析(DGA)数据,提出了基于CDAEN的变压器故障诊断方法。所提方法利用大量无标签样本进行预训练,优化模型参数,并利用少量有标签样本进行微调。实例分析表明,与基于反向传播神经网络(BPNN)、支持向量机(SVM)的故障诊断方法相比,所提方法的诊断正确率更高。

A CDAEN（Classified Deep Auto-Encoder Network） model is built based on the DAEN（Deep AutoEncoder Network）. Combined with the on-line monitored data of DGA（Dissolved Gas-in-oil Analysis） for power transformer,a method of transformer fault diagnosis based on CDEAN is proposed,which optimizes the parameters of CDAEN model by the pre-training with massive unlabeled samples and adjusts them with a few labeled samples. Results of case analysis show that the proposed method has higher diagnosis accuracy than those based on the BPNN（Back Propagation Neural Network） and the SVM（Support Vector Machine）.