变压器的故障预测研究  

Study on Fault Prediction of Transformer

作　　者：王磊[1] 陈长征[1] WANG Lei;CHEN Chang-zheng(Institute of Mechanical Engineering,Shenyang University of Technology,Liaoning Shenyang 110870,China)

机构地区：[1]沈阳工业大学机械工程学院,辽宁沈阳110870

出　　处：《机械设计与制造》2023年第1期65-68,73,共5页Machinery Design & Manufacture

基　　金：国家自然科学基金资助项目(51675350)。

摘　　要：针对变压器故障不能被及时发现的问题,提出了一种基于LSTM的变压器故障预测方法,用于提前预测变压器的故障趋势。该方法将变压器油中溶解的7种故障气体浓度作为网络的训练数据,采用不同网络结构建立了不同的变压器故障预测模型,最后,确立了最优的变压器故障预测模型。利用相同的变压器数据样本,对基于LSTM建立的变压器故障预测模型与基于灰色多变量方法建立的变压器故障预测模型进行测试,结果表明基于LSTM建立的变压器故障预测模型是非常有效的故障预测方法。Aiming at the problem that the transformer fault can not be found in time,a transformer fault prediction method based on LSTM is proposed to predict the fault trend of transformers in advance.this method takes the 7 fault gas concentrations dissolved in transformer oil as the training data of the network.different transformer fault prediction models are established by different network structures.Finally,the optimal transformer fault prediction model is established.By the same transformer data samples,the transformer fault prediction model based on LSTM and the transformer fault prediction model based on grey multivariable method are tested,and the results show that the fault prediction model based on LSTM is very effective.

关 键 词：变压器 故障趋势 预测 LSTM 故障气体 灰色多变量 

分 类 号：TH16[机械工程—机械制造及自动化] TH17